Stack: Difference between revisions

167,979 bytes added ,  1 month ago
m
(J)
 
(330 intermediate revisions by more than 100 users not shown)
Line 1:
[[Category:Encyclopedia]]{{task|Data Structures}}{{data structure}}[[Category:Classic CS problems and programs]]
[[Category:Encyclopedia]]
A '''stack''' is a container of elements with last in, first out access policy. Sometimes it also called '''LIFO'''. The stack is accessed through its '''top'''. The basic stack operations are:
{{data structure}}[[Category:Classic CS problems and programs]]
 
A '''stack''' is a container of elements with &nbsp; <big><u>l</u>ast <u>i</u>n, <u>f</u>irst <u>o</u>ut</big> &nbsp; access policy. &nbsp; Sometimes it also called '''LIFO'''.
* ''push'' stores a new element onto the stack top;
* ''pop'' returns the last pushed stack element, while removing it from the stack;
* ''empty'' tests if the stack contains no elements.
 
The stack is accessed through its '''top'''.
 
The basic stack operations are:
 
* &nbsp; ''push'' &nbsp; stores a new element onto the stack top;
* &nbsp; ''pop'' &nbsp; returns the last pushed stack element, while removing it from the stack;
* &nbsp; ''empty'' &nbsp; tests if the stack contains no elements.
 
<br>
Sometimes the last pushed stack element is made accessible for immutable access (for read) or mutable access (for write):
 
* &nbsp; ''top'' &nbsp; (sometimes called ''peek'' to keep with the ''p'' theme) returns the topmost element without modifying the stack.
 
<br>
Stacks as a containers presume copyable elements. I.e. stack elements have by-value semantics. This means that when an element is pushed onto the stack, a new instance of the element's type is created. This instance has a value equivalent to one the pushed element.
Stacks allow a very simple hardware implementation.
They are common in almost all processors.
In programming, stacks are also very popular for their way ('''LIFO''') of resource management, usually memory.
 
Nested scopes of language objects are naturally implemented by a stack (sometimes by multiple stacks).
Stacks allow a very simple hardware implementation. They are common in almost all processors. In programming stacks are also very popular for their way ('''LIFO''') of resource management, usually memory. Nested scopes of language objects are naturally implemented by a stack (sometimes by multiple stacks). This is a classical way to implement local variables of a reentrant or recursive subprogram. Stacks are also used to describe a formal computational framework. See [[wp:Stack_automaton|stack machine]]. Many algorithms in pattern matching, compiler construction (e.g. [[wp:Recursive_descent|recursive descent parsers]]), and machine learning (e.g. based on [[wp:Tree_traversal|tree traversal]]) have a natural representation in terms of stacks.
 
This is a classical way to implement local variables of a re-entrant or recursive subprogram. Stacks are also used to describe a formal computational framework.
 
See [[wp:Stack_automaton|stack machine]].
 
Many algorithms in pattern matching, compiler construction (e.g. [[wp:Recursive_descent|recursive descent parsers]]), and machine learning (e.g. based on [[wp:Tree_traversal|tree traversal]]) have a natural representation in terms of stacks.
 
 
;Task:
Create a stack supporting the basic operations: push, pop, empty.
 
 
{{Template:See also lists}}
<br><br>
 
=={{header|11l}}==
{{trans|Crystal}}
 
<syntaxhighlight lang="11l">[Int] stack
 
L(i) 1..10
stack.append(i)
 
L 10
print(stack.pop())</syntaxhighlight>
 
{{out}}
<pre>
10
9
8
7
6
5
4
3
2
1
</pre>
=={{header|6502 Assembly}}==
The 6502 has a built-in stack, which is located at memory addresses $0100-$01FF. The first thing most boot ROMs will do is set the stack to equal $FF. Only the X register can interact with the stack pointer's value directly, and it does so using <code>TSX</code> (transfer stack to X) and <code>TXS</code> (transfer X to stack.) Each push will decrement S by 1 and write that byte to the stack memory. On the original 6502, only the accumulator could be pushed to the stack, so programs running on those CPUs would often use sequences such as <code>TXA PHA</code> and <code>TYA PHA</code> to save the X and Y registers. This had the nasty habit of destroying the accumulator, which made saving these registers difficult. Fortunately, the 65c02 and its later revisions can push/pop X and Y directly without having to go through the accumulator first.
 
Push:
<syntaxhighlight lang="6502asm">PHA</syntaxhighlight>
Pop:
<syntaxhighlight lang="6502asm">PLA</syntaxhighlight>
Empty:
<syntaxhighlight lang="6502asm">TSX
CPX $FF
BEQ stackEmpty</syntaxhighlight>
Peek:
<syntaxhighlight lang="6502asm">TSX
LDA $0101,x</syntaxhighlight>
 
=={{header|68000 Assembly}}==
The 68000 is well-suited to stack data structures. Register A7 contains the stack pointer, however any address register can be used for a similar purpose. Any register from A0-A6 can be pointed to work RAM and used as a stack.
 
===Push===
You can push the contents of one or more variables.
<syntaxhighlight lang="68000devpac">LEA userStack,A0 ;initialize the user stack, points to a memory address in user RAM. Only do this once!
MOVEM.L D0-D3,-(A0) ;moves the full 32 bits of registers D0,D1,D2,D3 into the address pointed by A0, with pre-decrement</syntaxhighlight>
 
Unlike the "true" stack (A7), you can push a single byte onto the user stack and it won't get automatically padded with a trailing null byte.
 
===Pop===
The pop is just a reverse push.
<syntaxhighlight lang="68000devpac">MOVEM.L (A0)+,D0-D3 ;returns the four longs stored in the stack back to where they came from.</syntaxhighlight>
 
===Empty===
The stack is empty if and only if the stack pointer equals its initialized value. This is only true provided you have never adjusted the stack pointer except by pushing and popping.
<syntaxhighlight lang="68000devpac">CMPA.L #userStack,A0
BEQ StackIsEmpty</syntaxhighlight>
 
===Manually adjusting the stack===
You can offset the user stack (and the real stack) as follows:
<syntaxhighlight lang="68000devpac">LEA (4,SP),SP ;does the same thing to the stack as popping 4 bytes, except those bytes are not retrieved.</syntaxhighlight>
 
===Peek===
If you know the intended length of the last item on the stack (1, 2, or 4 bytes), you can load it into memory without popping it. This applies to both the real stack and a user stack you may have created. Since this operation doesn't alter the value of the stack pointer, you don't have to worry about misaligning the stack, but the value you peek at should be of the correct size or you'll be "peeking" at more than one item at the same time.
 
<syntaxhighlight lang="68000devpac">MOVE.W (SP),D0 ;load the top two bytes of the stack into D0
MOVE.W (A0),D0 ;load the top two bytes of A0 into D0</syntaxhighlight>
 
=={{header|8086 Assembly}}==
The 8086's hardware stack is very similar to that of [[Z80 Assembly]]. This is no coincidence, as the Z80 was based on the predecessor to the 8086.
 
<syntaxhighlight lang="asm">push ax ;push ax onto the stack
pop ax ; pop the top two bytes of the stack into ax</syntaxhighlight>
 
The "high" byte is pushed first, then the low byte. Popping does the opposite.
 
Depending on your assembler, the stack's initial value may be set using the <code>.stack</code> directive.
 
Like the Z80, the 8086 can only push or pop 2 bytes at a time. It's not possible to push <code>AH</code> without pushing <code>AL</code> alongside it. The stack can be used to exchange values of registers that even the <code>XCHG</code> command can't work with. This is done by deliberately pushing two registers and popping them in the "wrong" order.
 
The easiest way to "peek" is to pop then push that same register again.
<syntaxhighlight lang="asm">;get the top item of the stack
pop ax
push ax</syntaxhighlight>
 
The stack need not be accessed using these push and pop commands, it can also be read like any other area of memory. This is actually how [[C]] programs store and recall local variables and function arguments.
 
 
=={{header|ABAP}}==
 
This works for ABAP Version 7.40 and above
 
<syntaxhighlight lang="abap">
report z_stack.
 
interface stack.
methods:
push
importing
new_element type any
returning
value(new_stack) type ref to stack,
 
pop
exporting
top_element type any
returning
value(new_stack) type ref to stack,
 
empty
returning
value(is_empty) type abap_bool,
 
peek
exporting
top_element type any,
 
get_size
returning
value(size) type int4,
 
stringify
returning
value(stringified_stack) type string.
endinterface.
 
 
class character_stack definition.
public section.
interfaces:
stack.
 
 
methods:
constructor
importing
characters type string optional.
 
 
private section.
data:
characters type string.
endclass.
 
 
class character_stack implementation.
method stack~push.
characters = |{ new_element }{ characters }|.
 
new_stack = me.
endmethod.
 
 
method stack~pop.
if not me->stack~empty( ).
top_element = me->characters(1).
 
me->characters = me->characters+1.
endif.
 
new_stack = me.
endmethod.
 
 
method stack~empty.
is_empty = xsdbool( strlen( me->characters ) eq 0 ).
endmethod.
 
 
method stack~peek.
check not me->stack~empty( ).
 
top_element = me->characters(1).
endmethod.
 
 
method stack~get_size.
size = strlen( me->characters ).
endmethod.
 
 
method stack~stringify.
stringified_stack = cond string(
when me->stack~empty( )
then `empty`
else me->characters ).
endmethod.
 
 
method constructor.
check characters is not initial.
 
me->characters = characters.
endmethod.
endclass.
 
 
class integer_stack definition.
public section.
interfaces:
stack.
 
 
methods:
constructor
importing
integers type int4_table optional.
 
 
private section.
data:
integers type int4_table.
endclass.
 
 
class integer_stack implementation.
method stack~push.
append new_element to me->integers.
 
new_stack = me.
endmethod.
 
 
method stack~pop.
if not me->stack~empty( ).
top_element = me->integers[ me->stack~get_size( ) ].
 
delete me->integers index me->stack~get_size( ).
endif.
 
new_stack = me.
endmethod.
 
 
method stack~empty.
is_empty = xsdbool( lines( me->integers ) eq 0 ).
endmethod.
 
 
method stack~peek.
check not me->stack~empty( ).
 
top_element = me->integers[ lines( me->integers ) ].
endmethod.
 
 
method stack~get_size.
size = lines( me->integers ).
endmethod.
 
 
method stack~stringify.
stringified_stack = cond string(
when me->stack~empty( )
then `empty`
else reduce string(
init stack = ``
for integer in me->integers
next stack = |{ integer }{ stack }| ) ).
endmethod.
 
 
method constructor.
check integers is not initial.
 
me->integers = integers.
endmethod.
endclass.
 
 
start-of-selection.
data:
stack1 type ref to stack,
stack2 type ref to stack,
stack3 type ref to stack,
 
top_character type char1,
top_integer type int4.
 
stack1 = new character_stack( ).
stack2 = new integer_stack( ).
stack3 = new integer_stack( ).
 
write: |Stack1 = { stack1->stringify( ) }|, /.
stack1->push( 'a' )->push( 'b' )->push( 'c' )->push( 'd' ).
write: |push a, push b, push c, push d -> Stack1 = { stack1->stringify( ) }|, /.
stack1->pop( )->pop( importing top_element = top_character ).
write: |pop, pop and return element -> { top_character }, Stack1 = { stack1->stringify( ) }|, /, /.
 
write: |Stack2 = { stack2->stringify( ) }|, /.
stack2->push( 1 )->push( 2 )->push( 3 )->push( 4 ).
write: |push 1, push 2, push 3, push 4 -> Stack2 = { stack2->stringify( ) }|, /.
stack2->pop( )->pop( importing top_element = top_integer ).
write: |pop, pop and return element -> { top_integer }, Stack2 = { stack2->stringify( ) }|, /, /.
 
write: |Stack3 = { stack3->stringify( ) }|, /.
stack3->pop( ).
write: |pop -> Stack3 = { stack3->stringify( ) }|, /, /.
</syntaxhighlight>
 
{{out}}
<pre>
Stack1 = empty
 
push a, push b, push c, push d -> Stack1 = dcba
 
pop, pop and return element -> c, Stack1 = ba
 
 
Stack2 = empty
 
push 1, push 2, push 3, push 4 -> Stack2 = 4321
 
pop, pop and return element -> 3, Stack2 = 21
 
 
Stack3 = empty
 
pop -> Stack3 = empty
</pre>
 
=={{header|Action!}}==
===Static memory===
<syntaxhighlight lang="action!">DEFINE MAXSIZE="200"
BYTE ARRAY stack(MAXSIZE)
BYTE stacksize=[0]
 
BYTE FUNC IsEmpty()
IF stacksize=0 THEN
RETURN (1)
FI
RETURN (0)
 
PROC Push(BYTE v)
IF stacksize=maxsize THEN
PrintE("Error: stack is full!")
Break()
FI
stack(stacksize)=v
stacksize==+1
RETURN
 
BYTE FUNC Pop()
IF IsEmpty() THEN
PrintE("Error: stack is empty!")
Break()
FI
stacksize==-1
RETURN (stack(stacksize))
 
PROC TestIsEmpty()
IF IsEmpty() THEN
PrintE("Stack is empty")
ELSE
PrintE("Stack is not empty")
FI
RETURN
 
PROC TestPush(BYTE v)
PrintF("Push: %B%E",v)
Push(v)
RETURN
 
PROC TestPop()
BYTE v
 
Print("Pop: ")
v=Pop()
PrintBE(v)
RETURN
 
PROC Main()
TestIsEmpty()
TestPush(10)
TestIsEmpty()
TestPush(31)
TestPop()
TestIsEmpty()
TestPush(5)
TestPop()
TestPop()
TestPop()
RETURN</syntaxhighlight>
 
===Dynamic memory===
The user must type in the monitor the following command after compilation and before running the program!<pre>SET EndProg=*</pre>
{{libheader|Action! Tool Kit}}
<syntaxhighlight lang="action!">CARD EndProg ;required for ALLOCATE.ACT
 
INCLUDE "D2:ALLOCATE.ACT" ;from the Action! Tool Kit. You must type 'SET EndProg=*' from the monitor after compiling, but before running this program!
 
DEFINE PTR="CARD"
DEFINE NODE_SIZE="3"
TYPE StackNode=[BYTE data PTR nxt]
 
StackNode POINTER stack
 
BYTE FUNC IsEmpty()
IF stack=0 THEN
RETURN (1)
FI
RETURN (0)
 
PROC Push(BYTE v)
StackNode POINTER node
 
node=Alloc(NODE_SIZE)
node.data=v
node.nxt=stack
stack=node
RETURN
 
BYTE FUNC Pop()
StackNode POINTER node
BYTE v
IF IsEmpty() THEN
PrintE("Error stack is empty!")
Break()
FI
 
node=stack
v=node.data
stack=node.nxt
Free(node,NODE_SIZE)
RETURN (v)
 
PROC TestIsEmpty()
IF IsEmpty() THEN
PrintE("Stack is empty")
ELSE
PrintE("Stack is not empty")
FI
RETURN
 
PROC TestPush(BYTE v)
PrintF("Push: %B%E",v)
Push(v)
RETURN
 
PROC TestPop()
BYTE v
 
Print("Pop: ")
v=Pop()
PrintBE(v)
RETURN
 
PROC Main()
AllocInit(0)
stack=0
 
Put(125) PutE() ;clear screen
 
TestIsEmpty()
TestPush(10)
TestIsEmpty()
TestPush(31)
TestPop()
TestIsEmpty()
TestPush(5)
TestPop()
TestPop()
TestPop()
RETURN</syntaxhighlight>
{{out}}
Error at the end of program is intentional.
 
[https://gitlab.com/amarok8bit/action-rosetta-code/-/raw/master/images/Stack_array.png Screenshot from Atari 8-bit computer]
<pre>
Stack is empty
Push: 10
Stack is not empty
Push: 31
Pop: 31
Stack is not empty
Push: 5
Pop: 5
Pop: 10
Pop: Error: stack is empty!
 
RETURN
Error: 128
</pre>
 
=={{header|ActionScript}}==
In ActionScript an Array object provides stack functionality.
<langsyntaxhighlight lang="actionscript">var stack:Array = new Array();
var stack:Array = new Array();
stack.push(1);
stack.push(2);
trace(stack.pop()); // outputs "2"
trace(stack.pop()); // outputs "1"</syntaxhighlight>
</lang>
 
=={{header|Ada}}==
 
This is a generic stack implementation.
<syntaxhighlight lang="ada">generic
<lang ada>
type Element_Type is private;
generic
package Generic_Stack is
type Element_Type is private;
type Stack is private;
package Generic_Stack is
procedure Push (Item : Element_Type; Onto : in out Stack);
type Stack is private;
procedure PushPop (Item : out Element_Type; OntoFrom : in out Stack);
function Create return Stack;
procedure Pop (Item : out Element_Type; From : in out Stack);
Stack_Empty_Error : exception;
function Create return Stack;
private
Stack_Empty_Error : exception;
type Node;
private
type Stack is access Node;
type StackNode is access Node;record
Element : Element_Type;
type Node is record
Next : Stack Element := Element_Typenull;
end record;
Next : Stack := null;
end Generic_Stack;</syntaxhighlight>
end record;
<syntaxhighlight lang="ada">with Ada.Unchecked_Deallocation;
end Generic_Stack;
 
</lang>
package body Generic_Stack is
<lang ada>
with Ada.Unchecked_Deallocation;
------------
-- Create --
package body Generic_Stack is
------------
------------
--function Create --return Stack is
begin
------------
return (null);
functionend Create return Stack is;
 
begin
----------
return (null);
-- endPush Create;--
----------
 
----------
procedure Push(Item : Element_Type; Onto : in out Stack) is
-- Push --
Temp : Stack := new Node;
----------
begin
Temp.Element := Item;
procedure Push(Item : Element_Type; Onto : in out Stack) is
Temp : Stack.Next := new NodeOnto;
Onto := Temp;
begin
end Push;
Temp.Element := Item;
 
Temp.Next := Onto;
---------
Onto := Temp;
-- endPop Push;--
---------
 
---------
procedure Pop(Item : out Element_Type; From : in out Stack) is
-- Pop --
procedure Free is new Ada.Unchecked_Deallocation(Node, Stack);
---------
Temp : Stack := From;
begin
procedure Pop(Item : out Element_Type; From : in out Stack) is
if Temp = null then
procedure Free is new Ada.Unchecked_Deallocation(Node, Stack);
Temp : Stackraise := FromStack_Empty_Error;
begin end if;
Item if:= Temp = null then.Element;
From := raise Stack_Empty_ErrorTemp.Next;
end ifFree(Temp);
end Pop;
Item := Temp.Element;
 
From := Temp.Next;
end Generic_Stack;</syntaxhighlight>
Free(Temp);
end Pop;
end Generic_Stack;
</lang>
 
=={{header|ALGOL 68}}==
===ALGOL 68: Using linked list===
ALGOL 68 uses "HEAP" variables for new LINKs in a linked list. Generally ALGOL 68's
[[garbage collection|garbage collector]] should recover the LINK memory some time after a value is popped.
{{works with|ALGOL 68|Revision 1 - one extension to language used - PRAGMA READ - a non standard feature similar to C's #include directive.}}
<lang cpp>
{{works with|ALGOL 68G|Any - tested with release [http://sourceforge.net/projects/algol68/files/algol68g/algol68g-2.7 algol68g-2.7].}}
MODE VALUE = STRING; # type of a LINK in this STACK #
{{wont work 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] - due to extensive use of '''format'''[ted] ''transput''.}}
'''File: prelude/next_link.a68'''<syntaxhighlight lang="algol68"># -*- coding: utf-8 -*- #
CO REQUIRES:
MODE OBJVALUE = ~ # Mode/type of actual obj to be stacked #
END CO
 
MODE LINKOBJNEXTLINK = STRUCT(VALUE value, REF LINK next);
REF OBJNEXTLINK next,
MODE STACK = STRUCT(REF LINK first);
OBJVALUE value # ... etc. required #
);
 
PROC obj nextlink new = REF OBJNEXTLINK:
STRUCT (
HEAP OBJNEXTLINK;
PROC (REF STACK)VOID init,
 
PROC (REF STACK)BOOL non zero,
PROC obj nextlink free PROC= (REF STACK,OBJNEXTLINK VALUEfree)VOID append,:
next OF free := obj stack empty # give the garbage collector a BIG hint #</syntaxhighlight>'''File: prelude/stack_base.a68'''<syntaxhighlight lang="algol68"># -*- coding: utf-8 -*- #
PROC (REF STACK)VALUE pop,
CO REQUIRES:
PROC (REF STACK)STRING repr,
MODE OBJNEXTLINK = STRUCT(
PROC (REF STACK, STRING)BOOL index error mended
REF OBJNEXTLINK next,
) class stack = (
OBJVALUE value
# PROC init = # (REF STACK self)VOID:
);
first OF self := NIL,
# PROC nonobj zeronextlink new = # (REF STACKOBJNEXTLINK: self)BOOL:~,
PROC obj nextlink free = (REF LINK(firstOBJNEXTLINK OF selffree)VOID: ISNT NIL ,~
END CO
# PROC append = # (REF STACK self, VALUE value)VOID:
 
first OF self := HEAP LINK := (value, first OF self),
# actually a pointer to the last LINK, there ITEMs are ADDED, pushed & popped #
# PROC pop = # (REF STACK self)VALUE: (
MODE OBJSTACK = REF OBJNEXTLINK;
IF first OF self IS NIL THEN
 
STRING message = "pop from empty stack";
OBJSTACK obj stack empty = NIL;
IF NOT (index error mended OF class stack)(self, message) THEN
 
raise index error(message)
BOOL obj stack par = FALSE; # make code thread safe #
FI
SEMA obj stack sema = LEVEL ABS obj stack par;
FI;
# Warning: 1 SEMA for all stacks of type obj, i.e. not 1 SEMA per stack #
VALUE out = value OF first OF self;
 
first OF self := next OF first OF self;
PROC obj stack init = (REF OBJSTACK self)REF OBJSTACK:
out
self := obj stack empty;
),
 
# PROC repr = # (REF STACK self)STRING: (
# see if the program/coder wants the OBJ problem mended... #
STRING out := "(",
PROC (REF OBJSTACK #self#)BOOL obj stack index error mended
sep := "";
:= (REF OBJSTACK self)BOOL: (abend("obj stack index error"); stop);
REF LINK this := first OF self;
 
WHILE REF LINK(this) ISNT NIL DO
PROC on obj stack index error = (REF OBJSTACK self, PROC(REF OBJSTACK #self#)BOOL mended)VOID:
out +:= sep + """" + value OF this + """";
obj stack index error sepmended := ", "mended;
 
this := next OF this
PROC obj stack push = (REF OBJSTACK self, OBJVALUE obj)REF OBJSTACK:(
OD;
IF obj stack par THEN DOWN obj stack sema FI;
out+")"
self := obj nextlink new := (self, obj);
),
IF obj stack par THEN UP obj stack sema FI;
# PROC index error mended = # (REF STACK self, STRING message)BOOL:
self
FALSE # no mend applied #
);
 
# aliases: define a useful put (+=:) operator... #
PROC raise index error := (STRING message)VOID: stop;
OP +=: = (OBJVALUE obj, REF OBJSTACK self)REF OBJSTACK: obj stack push(self, obj);
 
PROC obj stack pop = (REF OBJSTACK self)OBJVALUE: (
STACK stack; (init OF class stack)(stack);
# DOWN obj stack sema; #
IF self IS obj stack empty THEN
IF NOT obj stack index error mended(self) THEN abend("obj stack index error") FI FI;
 
OBJNEXTLINK old head := self;
[]STRING sample = ("Was", "it", "a", "cat", "I", "saw");
OBJSTACK new head := next OF self;
OBJVALUE out := value OF old head;
obj nextlink free(old head); # freeing nextlink, NOT queue! #
self := new head;
#;UP obj stack sema; #
out
);
 
PROC obj stack is empty = (REF OBJSTACK self)BOOL:
FOR i TO UPB sample DO
(appendself OFIS classobj stack)(stack, sample[i])empty;
 
SKIP</syntaxhighlight>'''File: test/data_stigler_diet.a68'''<syntaxhighlight lang="algol68"># -*- coding: utf-8 -*- #
MODE DIETITEM = STRUCT(
STRING food, annual quantity, units, REAL cost
);
 
# Stigler's 1939 Diet ... #
FORMAT diet item fmt = $g": "g" "g" = $"zd.dd$;
[]DIETITEM stigler diet = (
("Cabbage", "111","lb.", 4.11),
("Dried Navy Beans", "285","lb.", 16.80),
("Evaporated Milk", "57","cans", 3.84),
("Spinach", "23","lb.", 1.85),
("Wheat Flour", "370","lb.", 13.33),
("Total Annual Cost", "","", 39.93)
)</syntaxhighlight>'''File: test/stack.a68'''<syntaxhighlight lang="algol68">#!/usr/bin/a68g --script #
# -*- coding: utf-8 -*- #
 
MODE OBJVALUE = DIETITEM;
PR read "prelude/next_link.a68" PR;
PR read "prelude/stack_base.a68" PR;
 
PR read "test/data_stigler_diet.a68" PR;
OBJSTACK example stack; obj stack init(example stack);
 
FOR i TO UPB stigler diet DO
# obj stack push(example stack, stigler diet[i]) #
stigler diet[i] +=: example stack
OD;
 
printf($"Items popped in reverse:"l$);
print(((repr OF class stack)(stack), newline))
WHILE NOT obj stack is empty(example stack) DO
</lang>
# OR example stack ISNT obj stack empty #
Output:
printf((diet item fmt, obj stack pop(example stack), $l$))
OD</syntaxhighlight>
{{out}}
<pre>
Items popped in reverse:
("saw", "I", "cat", "a", "it", "Was")
Total Annual Cost: = $39.93
Wheat Flour: 370 lb. = $13.33
Spinach: 23 lb. = $ 1.85
Evaporated Milk: 57 cans = $ 3.84
Dried Navy Beans: 285 lb. = $16.80
Cabbage: 111 lb. = $ 4.11
</pre>
'''See also:''' [[Queue#ALGOL_68|Queue]]
===ALGOL 68: Using FLEX array===
An alternative to using a linked list is to use a FLEX array.
<syntaxhighlight lang="algol68">
MODE DIETITEM = STRUCT (
STRING food, annual quantity, units, REAL cost
);
MODE OBJVALUE = DIETITEM;
 
# PUSH element to stack #
OP +:= = (REF FLEX[]OBJVALUE stack, OBJVALUE item) VOID:
BEGIN
FLEX[UPB stack + 1]OBJVALUE newstack;
newstack[2:UPB newstack] := stack;
newstack[1] := item;
stack := newstack
END;
 
OP POP = (REF FLEX[]OBJVALUE stack) OBJVALUE:
IF UPB stack > 0 THEN
OBJVALUE result = stack[1];
stack := stack[2:UPB stack];
result
ELSE
# raise index error; # SKIP
FI;
 
# Stigler's 1939 Diet ... #
FORMAT diet item fmt = $g": "g" "g" = $"zd.dd$;
[]DIETITEM stigler diet = (
("Cabbage", "111","lb.", 4.11),
("Dried Navy Beans", "285","lb.", 16.80),
("Evaporated Milk", "57","cans", 3.84),
("Spinach", "23","lb.", 1.85),
("Wheat Flour", "370","lb.", 13.33),
("Total Annual Cost", "","", 39.93)
);
 
FLEX[0]DIETITEM example stack;
FOR i TO UPB stigler diet DO
example stack +:= stigler diet[i]
OD;
printf($"Items popped in reverse:"l$);
WHILE UPB example stack > 0 DO
printf((diet item fmt, POP example stack, $l$))
OD
</syntaxhighlight>
{{out}}
<pre>
Items popped in reverse:
Total Annual Cost: = $39.93
Wheat Flour: 370 lb. = $13.33
Spinach: 23 lb. = $ 1.85
Evaporated Milk: 57 cans = $ 3.84
Dried Navy Beans: 285 lb. = $16.80
Cabbage: 111 lb. = $ 4.11
</pre>
 
=={{header|ALGOL W}}==
<syntaxhighlight lang="algolw">begin
% define a Stack type that will hold StringStackElements %
% and the StringStackElement type %
% we would need separate types for other element types %
record StringStack ( reference(StringStackElement) top );
record StringStackElement ( string(8) element
; reference(StringStackElement) next
);
% adds e to the end of the StringStack s %
procedure pushString ( reference(StringStack) value s
; string(8) value e
) ;
top(s) := StringStackElement( e, top(s) );
% removes and returns the top element from the StringStack s %
% asserts the Stack is not empty, which will stop the %
% program if it is %
string(8) procedure popString ( reference(StringStack) value s ) ;
begin
string(8) v;
assert( not isEmptyStringStack( s ) );
v := element(top(s));
top(s):= next(top(s));
v
end popStringStack ;
% returns the top element of the StringStack s %
% asserts the Stack is not empty, which will stop the %
% program if it is %
string(8) procedure peekStringStack ( reference(StringStack) value s ) ;
begin
assert( not isEmptyStringStack( s ) );
element(top(s))
end popStringStack ;
% returns true if the StringStack s is empty, false otherwise %
logical procedure isEmptyStringStack ( reference(StringStack) value s ) ; top(s) = null;
begin % test the StringStack operations %
reference(StringStack) s;
s := StringStack( null );
pushString( s, "up" );
pushString( s, "down" );
pushString( s, "strange" );
pushString( s, "charm" );
while not isEmptyStringStack( s ) do write( popString( s )
, if isEmptyStringStack( s ) then "(empty)"
else peekStringStack( s )
)
end
end.</syntaxhighlight>
{{out}}
<pre>
charm strange
strange down
down up
up (empty)
</pre>
 
=={{header|Applesoft BASIC}}==
<syntaxhighlight lang="basic">100 DIM STACK$(1000)
110 DATA "(2*A)","PI","","TO BE OR","NOT TO BE"
120 FOR I = 1 TO 5
130 READ ELEMENT$
140 GOSUB 500_PUSH
150 NEXT
200 GOSUB 400 POP AND PRINT
210 GOSUB 300_EMPTY AND PRINT
220 FOR I = 1 TO 4
230 GOSUB 400 POP AND PRINT
240 NEXT
250 GOSUB 300_EMPTY AND PRINT
260 END
300 GOSUB 700_EMPTY
310 PRINT "STACK IS ";
320 IF NOT EMPTY THEN PRINT "NOT ";
330 PRINT "EMPTY"
340 RETURN
400 GOSUB 600 POP
410 PRINT ELEMENT$
420 RETURN
500 REM
510 REM PUSH
520 REM
530 LET STACK$(SP) = ELEMENT$
540 LET SP = SP + 1
550 RETURN
600 REM
610 REM POP
620 REM
630 IF SP THEN SP = SP - 1
640 LET ELEMENT$ = STACK$(SP)
650 LET STACK$(SP) = ""
660 RETURN
700 REM
710 REM EMPTY
720 REM
730 LET EMPTY = SP = 0
740 RETURN
</syntaxhighlight>
 
{{out}}
<pre>NOT TO BE
STACK IS NOT EMPTY
TO BE OR
 
PI
(2*A)
STACK IS EMPTY
</pre>
 
=={{header|ARM Assembly}}==
The stack is held in register 13, or <code>r13</code> but more commonly referred to as <code>SP</code> for clarity.
 
Pushing and popping multiple values is very similar to [[68000 Assembly]].
<syntaxhighlight lang="arm assembly">STMFD sp!,{r0-r12,lr} ;push r0 thru r12 and the link register
LDMFD sp!,{r0-r12,pc} ;pop r0 thru r12, and the value that was in the link register is put into the program counter.
;This acts as a pop and return command all-in-one. (Most programs use bx lr to return.)</syntaxhighlight>
 
Like in 68000 Assembly, you are not limited to using <code>SP</code> as the source/destination for these commands; any register can fulfill that role. If you wish to have multiple stacks, then so be it.
 
The stack pointer will work with any operation the other registers can. As such, a peek can be done by using an <code>LDR</code> with the stack pointer as the address register:
 
<syntaxhighlight lang="arm assembly">LDR r0,[sp] ;load the top of the stack into r0</syntaxhighlight>
 
The order in which registers are pushed/popped is always the same, no matter which order you list the registers in your source code. If you want to push some registers and purposefully pop them into different registers, you'll need to push/pop them separately.
 
A check if the stack is empty is also very simple, provided the initial value of the stack pointer was saved at the start of the program, or (more likely) was loaded from a nearby memory location.
 
<syntaxhighlight lang="arm assembly">;this example uses VASM syntax which considers a "word" to be 16-bit regardless of the architecture
InitStackPointer: .long 0x3FFFFFFF ;other assemblers would call this a "word"
 
MOV R1,#InitStackPointer
LDR SP,[R1] ;set up the stack pointer
LDR R2,[R1] ;also load it into R2
;There's no point in checking since we haven't pushed/popped anything but just for demonstration purposes we'll check now
CMP SP,R2
BEQ StackIsEmpty</syntaxhighlight>
 
In THUMB mode, the <code>PUSH</code> and <code>POP</code> commands replace <code>STMFD</code> and <code>LDMFD</code>. They work in a similar fashion, but are limited to just the stack unlike the real <code>STMFD</code> and <code>LDMFD</code> commands which can use any register as the "stack pointer."
 
=={{header|Arturo}}==
<syntaxhighlight lang="rebol">Stack: $[]-> []
 
pushTo: function [st val]-> 'st ++ val
popStack: function [st] [
result: last st
remove 'st .index (size st)-1
return result
]
emptyStack: function [st]-> empty 'st
printStack: function [st]-> print st
 
st: new Stack
 
pushTo st "one"
pushTo st "two"
pushTo st "three"
printStack st
 
print popStack st
printStack st
 
emptyStack st
print ["finally:" st]</syntaxhighlight>
 
{{out}}
 
<pre>one two three
three
one two
finally: []</pre>
 
=={{header|ATS}}==
<syntaxhighlight lang="ats">(* Stacks implemented as linked lists. *)
 
(* A nonlinear stack type of size n, which is good for when you are
using a garbage collector or can let the memory leak. *)
typedef stack_t (t : t@ype+, n : int) = list (t, n)
typedef stack_t (t : t@ype+) = [n : int] stack_t (t, n)
 
(* A linear stack type of size n, which requires (and will enforce)
explicit freeing. (Note that a "peek" function for a linear stack
is a complicated topic. But the task avoids this issue.) *)
viewtypedef stack_vt (vt : vt@ype+, n : int) = list_vt (vt, n)
viewtypedef stack_vt (vt : vt@ype+) = [n : int] stack_vt (vt, n)
 
(* Proof that a given nonlinear stack does not have a nonnegative
size. *)
prfn
lemma_stack_t_param {n : int} {t : t@ype}
(stack : stack_t (t, n)) :<prf>
[0 <= n] void =
lemma_list_param stack
 
(* Proof that a given linear stack does not have a nonnegative
size. *)
prfn
lemma_stack_vt_param {n : int} {vt : vt@ype}
(stack : !stack_vt (vt, n)) :<prf>
[0 <= n] void =
lemma_list_vt_param stack
 
(* Create an empty nonlinear stack. *)
fn {}
stack_t_nil {t : t@ype} () :<> stack_t (t, 0) =
list_nil ()
 
(* Create an empty linear stack. *)
fn {}
stack_vt_nil {vt : vt@ype} () :<> stack_vt (vt, 0) =
list_vt_nil ()
 
(* Is a nonlinear stack empty? *)
fn {}
stack_t_is_empty {n : int} {t : t@ype}
(stack : stack_t (t, n)) :<>
[empty : bool | empty == (n == 0)]
bool empty =
case+ stack of
| list_nil _ => true
| list_cons _ => false
 
(* Is a linear stack empty? *)
fn {}
stack_vt_is_empty {n : int} {vt : vt@ype}
(* ! = pass by value; stack is preserved. *)
(stack : !stack_vt (vt, n)) :<>
[empty : bool | empty == (n == 0)]
bool empty =
case+ stack of
| list_vt_nil _ => true
| list_vt_cons _ => false
 
(* Push to a nonlinear stack that is stored in a variable. *)
fn {t : t@ype}
stack_t_push {n : int}
(stack : &stack_t (t, n) >> stack_t (t, m),
x : t) :<!wrt>
(* It is proved that the stack is raised one higher. *)
#[m : int | 1 <= m; m == n + 1]
void =
let
prval _ = lemma_stack_t_param stack
prval _ = prop_verify {0 <= n} ()
in
stack := list_cons (x, stack)
end
 
(* Push to a linear stack that is stored in a variable. Beware: if x
is linear, it is consumed. *)
fn {vt : vt@ype}
stack_vt_push {n : int}
(stack : &stack_vt (vt, n) >> stack_vt (vt, m),
x : vt) :<!wrt>
(* It is proved that the stack is raised one higher. *)
#[m : int | 1 <= m; m == n + 1]
void =
let
prval _ = lemma_stack_vt_param stack
prval _ = prop_verify {0 <= n} ()
in
stack := list_vt_cons (x, stack)
end
 
(* Pop from a nonlinear stack that is stored in a variable. It is
impossible (unless you cheat the typechecker) to pop from an empty
stack. *)
fn {t : t@ype}
stack_t_pop {n : int | 1 <= n}
(stack : &stack_t (t, n) >> stack_t (t, m)) :<!wrt>
(* It is proved that the stack is lowered by one. *)
#[m : int | m == n - 1]
t =
case+ stack of
| list_cons (x, tail) =>
begin
stack := tail;
x
end
 
(* Pop from a linear stack that is stored in a variable. It is
impossible (unless you cheat the typechecker) to pop from an empty
stack. *)
fn {vt : vt@ype}
stack_vt_pop {n : int | 1 <= n}
(stack : &stack_vt (vt, n) >> stack_vt (vt, m)) :<!wrt>
(* It is proved that the stack is lowered by one. *)
#[m : int | m == n - 1]
vt =
case+ stack of
| ~ list_vt_cons (x, tail) => (* ~ = the top node is consumed. *)
begin
stack := tail;
x
end
 
(* A linear stack has to be consumed. *)
extern fun {vt : vt@ype}
stack_vt_free$element_free (x : vt) :<> void
fn {vt : vt@ype}
stack_vt_free {n : int}
(stack : stack_vt (vt, n)) :<> void =
let
fun
loop {m : int | 0 <= m}
.<m>. (* <-- proof of loop termination *)
(stk : stack_vt (vt, m)) :<> void =
case+ stk of
| ~ list_vt_nil () => begin end
| ~ list_vt_cons (x, tail) =>
begin
stack_vt_free$element_free x;
loop tail
end
 
prval _ = lemma_stack_vt_param stack
in
loop stack
end
 
implement
main0 () =
let
var nonlinear_stack : stack_t (int) = stack_t_nil ()
var linear_stack : stack_vt (int) = stack_vt_nil ()
implement stack_vt_free$element_free<int> x = begin end
 
overload is_empty with stack_t_is_empty
overload is_empty with stack_vt_is_empty
 
overload push with stack_t_push
overload push with stack_vt_push
 
overload pop with stack_t_pop
overload pop with stack_vt_pop
in
println! ("nonlinear_stack is empty? ", is_empty nonlinear_stack);
println! ("linear_stack is empty? ", is_empty linear_stack);
 
println! ("pushing 3, 2, 1...");
push (nonlinear_stack, 3);
push (nonlinear_stack, 2);
push (nonlinear_stack, 1);
push (linear_stack, 3);
push (linear_stack, 2);
push (linear_stack, 1);
 
println! ("nonlinear_stack is empty? ", is_empty nonlinear_stack);
println! ("linear_stack is empty? ", is_empty linear_stack);
 
println! ("popping nonlinear_stack: ", (pop nonlinear_stack) : int);
println! ("popping nonlinear_stack: ", (pop nonlinear_stack) : int);
println! ("popping nonlinear_stack: ", (pop nonlinear_stack) : int);
 
println! ("popping linear_stack: ", (pop linear_stack) : int);
println! ("popping linear_stack: ", (pop linear_stack) : int);
println! ("popping linear_stack: ", (pop linear_stack) : int);
 
println! ("nonlinear_stack is empty? ", is_empty nonlinear_stack);
println! ("linear_stack is empty? ", is_empty linear_stack);
 
stack_vt_free<int> linear_stack
end</syntaxhighlight>
 
{{out}}
<pre>$ patscc -O2 -DATS_MEMALLOC_LIBC stack-postiats.dats && ./a.out
nonlinear_stack is empty? true
linear_stack is empty? true
pushing 3, 2, 1...
nonlinear_stack is empty? false
linear_stack is empty? false
popping nonlinear_stack: 1
popping nonlinear_stack: 2
popping nonlinear_stack: 3
popping linear_stack: 1
popping linear_stack: 2
popping linear_stack: 3
nonlinear_stack is empty? true
linear_stack is empty? true</pre>
 
=={{header|AutoHotkey}}==
<syntaxhighlight lang="autohotkey">msgbox % stack("push", 4)
<lang AutoHotkey>
msgbox % stack("push", 4)
msgbox % stack("push", 5)
msgbox % stack("peek")
Line 198 ⟶ 1,195:
return 0
}
}</syntaxhighlight>
 
=={{header|AWK}}==
<syntaxhighlight lang="awk">function deque(arr) {
arr["start"] = 0
arr["end"] = 0
}
 
</lang>
function dequelen(arr) {
return arr["end"] - arr["start"]
}
 
function empty(arr) {
return dequelen(arr) == 0
}
 
function push(arr, elem) {
arr[++arr["end"]] = elem
}
 
function pop(arr) {
if (empty(arr)) {
return
}
return arr[arr["end"]--]
}
 
function unshift(arr, elem) {
arr[arr["start"]--] = elem
}
 
function shift(arr) {
if (empty(arr)) {
return
}
return arr[++arr["start"]]
}
 
function peek(arr) {
if (empty(arr)) {
return
}
return arr[arr["end"]]
}
 
function printdeque(arr, i, sep) {
printf("[")
for (i = arr["start"] + 1; i <= arr["end"]; i++) {
printf("%s%s", sep, arr[i])
sep = ", "
}
printf("]\n")
}
 
BEGIN {
deque(q)
for (i = 1; i <= 10; i++) {
push(q, i)
}
printdeque(q)
for (i = 1; i <= 10; i++) {
print pop(q)
}
printdeque(q)
}</syntaxhighlight>
 
=={{header|Axe}}==
 
<syntaxhighlight lang="axe">0→S
Lbl PUSH
r₁→{L₁+S}ʳ
S+2→S
Return
 
Lbl POP
S-2→S
{L₁+S}ʳ
Return
 
Lbl EMPTY
S≤≤0
Return</syntaxhighlight>
 
=={{header|Babel}}==
<syntaxhighlight lang="babel">main :
{ (1 2 3) foo set -- foo = (1 2 3)
4 foo push -- foo = (1 2 3 4)
0 foo unshift -- foo = (0 1 2 3 4)
foo pop -- foo = (0 1 2 3)
foo shift -- foo = (1 2 3)
check_foo
{ foo pop } 4 times -- Pops too many times, but this is OK and Babel won't complain
check_foo }
 
empty? : nil? -- just aliases 'empty?' to the built-in operator 'nil?'
 
check_foo! :
{ "foo is "
{foo empty?) {nil} {"not " .} ifte
"empty" .
cr << }
</syntaxhighlight>
{{out}}
<pre>
foo is not empty
foo is empty</pre>
 
=={{header|Batch File}}==
This implementation uses an environment variable naming convention to implement a stack as a pseudo object containing a pseudo dynamic array and top attribute, as well as an empty "method" that is a sort of macro. The implementation depends on delayed expansion being enabled at the time of each call to a stack function. More complex variations can be written that remove this limitation.
<syntaxhighlight lang="dos">@echo off
setlocal enableDelayedExpansion
 
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: LIFO stack usage
 
:: Define the stack
call :newStack myStack
 
:: Push some values onto the stack
for %%A in (value1 value2 value3) do call :pushStack myStack %%A
 
:: Test if stack is empty by examining the top "attribute"
if myStack.top==0 (echo myStack is empty) else (echo myStack is NOT empty)
 
:: Peek at the top stack value
call:peekStack myStack val && echo a peek at the top of myStack shows !val!
 
:: Pop the top stack value
call :popStack myStack val && echo popped myStack value=!val!
 
:: Push some more values onto the stack
for %%A in (value4 value5 value6) do call :pushStack myStack %%A
 
:: Process the remainder of the stack
:processStack
call :popStack myStack val || goto :stackEmpty
echo popped myStack value=!val!
goto :processStack
:stackEmpty
 
:: Test if stack is empty using the empty "method"/"macro". Use of the
:: second IF statement serves to demonstrate the negation of the empty
:: "method". A single IF could have been used with an ELSE clause instead.
if %myStack.empty% echo myStack is empty
if not %myStack.empty% echo myStack is NOT empty
exit /b
 
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: LIFO stack definition
 
:newStack stackName
set /a %~1.top=0
:: Define an empty "method" for this stack as a sort of macro
set "%~1.empty=^!%~1.top^! == 0"
exit /b
 
:pushStack stackName value
set /a %~1.top+=1
set %~1.!%~1.top!=%2
exit /b
 
:popStack stackName returnVar
:: Sets errorlevel to 0 if success
:: Sets errorlevel to 1 if failure because stack was empty
if !%~1.top! equ 0 exit /b 1
for %%N in (!%~1.top!) do (
set %~2=!%~1.%%N!
set %~1.%%N=
)
set /a %~1.top-=1
exit /b 0
 
:peekStack stackName returnVar
:: Sets errorlevel to 0 if success
:: Sets errorlevel to 1 if failure because stack was empty
if !%~1.top! equ 0 exit /b 1
for %%N in (!%~1.top!) do set %~2=!%~1.%%N!
exit /b 0</syntaxhighlight>
 
=={{header|BASIC}}==
==={{header|BBC BASIC}}===
{{works with|BBC BASIC for Windows}}
<syntaxhighlight lang="bbcbasic"> STACKSIZE = 1000
FOR n = 3 TO 5
PRINT "Push ";n : PROCpush(n)
NEXT
PRINT "Pop " ; FNpop
PRINT "Push 6" : PROCpush(6)
REPEAT
PRINT "Pop " ; FNpop
UNTIL FNisempty
PRINT "Pop " ; FNpop
END
DEF PROCpush(n) : LOCAL f%
DEF FNpop : LOCAL f% : f% = 1
DEF FNisempty : LOCAL f% : f% = 2
PRIVATE stack(), sptr%
DIM stack(STACKSIZE-1)
CASE f% OF
WHEN 0:
IF sptr% = DIM(stack(),1) ERROR 100, "Error: stack overflowed"
stack(sptr%) = n
sptr% += 1
WHEN 1:
IF sptr% = 0 ERROR 101, "Error: stack empty"
sptr% -= 1
= stack(sptr%)
WHEN 2:
= (sptr% = 0)
ENDCASE
ENDPROC</syntaxhighlight>
{{out}}
<pre>
Push 3
Push 4
Push 5
Pop 5
Push 6
Pop 6
Pop 4
Pop 3
Pop
Error: stack empty
</pre>
 
=={{header|beeswax}}==
Beeswax is a stack-based language. The instruction pointers (bees) carry small local stacks (lstacks) of fixed length 3 that can interact with the global stack (gstack) of unrestricted length. The local stacks do not behave exactly like the stack specified in this task, but the global stack does.
 
'''Push (1)''': <code>f</code> pushes the topmost value of lstack on gstack.
<pre> instruction: _f
 
gstack: UInt64[0]• (at the beginning of a program lstack is initialized to [0 0 0]</pre>
 
'''Push (2)''': <code>e</code> pushes all three lstack values on gstack, in reversed order.
<pre> instruction: _e
 
gstack: UInt64[0 0 0]• (at the beginning of a program lstack is initialized to [0 0 0]</pre>
 
'''Push (3)''': <code>i</code> pushes an integer from STDIN as UInt64 value on gstack.
<pre> instruction: _i
input: i123
 
gstack: UInt64[123]•</pre>
 
'''Push (4)''': <code>c</code> pushes the Unicode codepoint value of a character from STDIN as UInt64 value on gstack.
<pre> instruction: _c
input: cH
 
gstack: UInt64[72]•</pre>
 
'''Push (5)''': <code>V</code> pushes the Unicode codepoint values of the characters of a string given at STDIN as UInt64 values on gstack, last character, followed by newline on top.
<pre> instruction: _V
input: sHello, α∀
 
gstack: UInt64[72 101 108 108 111 44 32 945 8704 10]•</pre>
 
'''Pop''': <code>g{?</code> reads the top value of gstack and stores it on top of lstack. Then outputs top value of lstack to STDOUT and finally pops gstack.
 
'''Empty''': <code>Ag?';`gstack is empty`</code> pushes length of gstack on gstack, reads top value of gstack, stores it as top value of lstack and prints <code>gstack is empty</code> if lstack top=0.
 
'''Top''': <code>g{</code> reads the top value of gstack, stores it on top of lstack. Then outputs top value of lstack to STDOUT. If gstack is empty, this instruction does not do anything but return the topmost value of lstack.
 
To make sure that there is any value on gstack, you would need to check for gstack length first, using the method shown in the “'''Empty'''” example above:
<pre>*Ag'{`gstack empty, no value to return`</pre>
This method returns the top value of gstack only if gstack is not empty, otherwise it outputs <code>gstack empty, no value to return</code> to STDOUT.
 
=={{header|BQN}}==
Representing the stack as an array, pushing is appending, popping is removing the last element, and checking emptiness is checking the length.
 
<syntaxhighlight lang="bqn"> Push ← ∾
Pop ← ¯1⊸↓
¯1⊸↓
Empty ← 0=≠
0=≠
1‿2‿3 Push 4
⟨ 1 2 3 4 ⟩
Pop 1‿2‿3
⟨ 1 2 ⟩
Empty 1‿2‿3
0
Empty ⟨⟩
1</syntaxhighlight>
 
=={{header|Bracmat}}==
A stack is easiest implemented as a dotted list <code>top.top-1.top-2.<i>[...]</i>.</code>. In the example below we also introduce a 'class' <code>stack</code>, instantiated in the 'object' <code>Stack</code>. The class has a member variable <code>S</code> and methods <code>push</code>,<code>pop</code>, <code>top</code> and <code>empty</code>. As a side note, <code>.</code> is to <code>..</code> as C's <code>.</code> is to <code>-&gt;</code>. In a method's body, <code>its</code> refers to the object itself. (Analogous to <code>(*this)</code> in C++.)
<syntaxhighlight lang="bracmat">( ( stack
= (S=)
(push=.(!arg.!(its.S)):?(its.S))
( pop
= top.!(its.S):(%?top.?(its.S))&!top
)
(top=top.!(its.S):(%?top.?)&!top)
(empty=.!(its.S):)
)
& new$stack:?Stack
& (Stack..push)$(2*a)
& (Stack..push)$pi
& (Stack..push)$
& (Stack..push)$"to be or"
& (Stack..push)$"not to be"
& out$((Stack..pop)$|"Cannot pop (a)")
& out$((Stack..top)$|"Cannot pop (b)")
& out$((Stack..pop)$|"Cannot pop (c)")
& out$((Stack..pop)$|"Cannot pop (d)")
& out$((Stack..pop)$|"Cannot pop (e)")
& out$((Stack..pop)$|"Cannot pop (f)")
& out$((Stack..pop)$|"Cannot pop (g)")
& out$((Stack..pop)$|"Cannot pop (h)")
& out
$ ( str
$ ( "Stack is "
((Stack..empty)$&|not)
" empty"
)
)
&
);</syntaxhighlight>
{{out}}
<pre>not to be
to be or
to be or
 
pi
2*a
Cannot pop (g)
Cannot pop (h)
Stack is empty</pre>
 
=={{header|Brat}}==
Built in arrays have push, pop, and empty? methods:
 
<syntaxhighlight lang="brat">stack = []
stack.push 1
stack.push 2
stack.push 3
 
until { stack.empty? } { p stack.pop }</syntaxhighlight>
 
{{out}}
<pre>3
2
1</pre>
 
=={{header|C}}==
Macro expanding to type flexible stack routines.
<lang c>
<syntaxhighlight lang="c">#include <stdio.h>
#include <stdlib.h>
 
/* to read expanded code, run through cpp | indent -st */
#define DECL_STACK_TYPE(type, name) \
typedef struct stk_##name##_t{type *buf; size_t alloc,len;}*stk_##name; \
stk_##name stk_##name##_create(size_t init_size) { \
stk_##name s; if (!init_size) init_size = 4; \
s = malloc(sizeof(struct stk_##name##_t)); \
if (!s) return 0; \
s->buf = malloc(sizeof(type) * init_size); \
if (!s->buf) { free(s); return 0; } \
s->len = 0, s->alloc = init_size; \
return s; } \
int stk_##name##_push(stk_##name s, type item) { \
type *tmp; \
if (s->len >= s->alloc) { \
tmp = realloc(s->buf, s->alloc*2*sizeof(type)); \
if (!tmp) return -1; s->buf = tmp; \
s->alloc *= 2; } \
s->buf[s->len++] = item; \
return s->len; } \
type stk_##name##_pop(stk_##name s) { \
type tmp; \
if (!s->len) abort(); \
tmp = s->buf[--s->len]; \
if (s->len * 2 <= s->alloc && s->alloc >= 8) { \
s->alloc /= 2; \
s->buf = realloc(s->buf, s->alloc * sizeof(type));} \
return tmp; } \
void stk_##name##_delete(stk_##name s) { \
free(s->buf); free(s); }
 
#define stk_empty(s) (!(s)->len)
#define stk_size(s) ((s)->len)
 
DECL_STACK_TYPE(int, int)
 
int main(void)
{
int i;
stk_int stk = stk_int_create(0);
 
printf("pushing: ");
for (i = 'a'; i <= 'z'; i++) {
printf(" %c", i);
stk_int_push(stk, i);
}
 
printf("\nsize now: %d", stk_size(stk));
printf("\nstack is%s empty\n", stk_empty(stk) ? "" : " not");
 
printf("\npoppoing:");
while (stk_size(stk))
printf(" %c", stk_int_pop(stk));
printf("\nsize now: %d", stk_size(stk));
printf("\nstack is%s empty\n", stk_empty(stk) ? "" : " not");
 
/* stk_int_pop(stk); <-- will abort() */
stk_int_delete(stk);
return 0;
}</syntaxhighlight>
===Or===
<syntaxhighlight lang="c">#include <stdio.h>
#include <stdlib.h>
#include <stddef.h>
Line 272 ⟶ 1,676:
s->bottom = realloc(s->bottom, new_size);
check_pointer(s->bottom);
s->allocated_top = s->bottom + qtty - 1;}</syntaxhighlight>
</lang>
 
=={{header|C sharp|C #}}==
<syntaxhighlight lang="csharp">// Non-Generic Stack
System.Collections.Stack stack = new System.Collections.Stack();
stack.Push( obj );
bool isEmpty = stack.Count == 0;
object top = stack.Peek(); // Peek without Popping.
top = stack.Pop();
 
// Non-Generic Stack
System.Collections.Generic.Stack<Foo> stack = new System.Collections.Generic.Stack<Foo>();
stack.Push(new obj Foo());
bool isEmpty = stack.Count == 0;
objectFoo top = stack.Peek(); // Peek without Popping.
top = stack.Pop();</syntaxhighlight>
 
// Generic Stack
System.Collections.Generic.Stack<Foo> stack = new System.Collections.Generic.Stack<Foo>();
stack.Push(new Foo());
bool isEmpty = stack.Count == 0;
Foo top = stack.Peek(); // Peek without Popping.
top = stack.Pop();
 
=={{header|C++}}==
 
{{libheader|STL}}
 
The C++ standard library already provides a ready-made stack class. You get it by writing
<syntaxhighlight lang="cpp">#include <stack></syntaxhighlight>
 
<lang cpp>
#include <stack>
</lang>
 
and then using the <tt>std::stack</tt> class.
 
An example of an explicit implementation of a stack class (which actually implements the standard stack class, except that the standard one is in namespace std):
<syntaxhighlight lang="cpp">#include <deque>
 
<lang cpp>
#include <deque>
template <class T, class Sequence = std::deque<T> >
class stack {
Line 360 ⟶ 1,754:
{
return !(x < y);
}</syntaxhighlight>
}
</lang>
 
== {{header|Common LispClojure}} ==
As is mentioned in the Common Lisp example below, built in cons-based lists work just fine. In this implementation, the list is wrapped in a datatype, providing a stateful solution.
<syntaxhighlight lang="lisp">(deftype Stack [elements])
 
(def stack (Stack (ref ())))
It's a bit unusual to write a wrapper for a stack in Common Lisp; built-in cons-based lists work just fine. Nonetheless, here's an implementation where the list is wrapped in a structure, providing a stateful solution.
 
<lang lisp>(defstructdefn push-stack
"Pushes an item to the top of the stack."
[x] (dosync (alter (:elements stack) conj x)))
 
(defn pop-stack
"Pops an item from the top of the stack."
[] (let [fst (first (deref (:elements stack)))]
(dosync (alter (:elements stack) rest)) fst))
 
(defn top-stack
"Shows what's on the top of the stack."
[] (first (deref (:elements stack))))
 
(defn empty-stack?
"Tests whether or not the stack is empty."
[] (= () (deref (:elements stack))))</syntaxhighlight>
 
We can make this a bit smaller and general by using defprotocol along with deftype. Here is a revised version using defprotocol.
 
<syntaxhighlight lang="lisp">(defprotocol StackOps
(push-stack [this x] "Pushes an item to the top of the stack.")
(pop-stack [this] "Pops an item from the top of the stack.")
(top-stack [this] "Shows what's on the top of the stack.")
(empty-stack? [this] "Tests whether or not the stack is empty."))
(deftype Stack [elements]
StackOps
(push-stack [x] (dosync (alter elements conj x)))
(pop-stack [] (let [fst (first (deref elements))]
(dosync (alter elements rest)) fst))
(top-stack [] (first (deref elements)))
(empty-stack? [] (= () (deref elements))))
 
(def stack (Stack (ref ())))</syntaxhighlight>
 
=={{header|CLU}}==
<syntaxhighlight lang="clu">% Stack
stack = cluster [T: type] is new, push, pop, peek, empty
rep = array[T]
new = proc () returns (cvt)
return (rep$new())
end new
empty = proc (s: cvt) returns (bool)
return (rep$size(s) = 0)
end empty;
push = proc (s: cvt, val: T)
rep$addh(s, val)
end push;
pop = proc (s: cvt) returns (T) signals (empty)
if rep$empty(s)
then signal empty
else return(rep$remh(s))
end
end pop
peek = proc (s: cvt) returns (T) signals (empty)
if rep$empty(s)
then signal empty
else return(s[rep$high(s)])
end
end peek
end stack
 
start_up = proc ()
po: stream := stream$primary_output()
% Make a stack
s: stack[int] := stack[int]$new()
% Push 1..10 onto the stack
for i: int in int$from_to(1, 10) do
stack[int]$push(s, i)
end
% Pop items off the stack until the stack is empty
while ~stack[int]$empty(s) do
stream$putl(po, int$unparse(stack[int]$pop(s)))
end
% Trying to pop off the stack now should raise 'empty'
begin
i: int := stack[int]$pop(s)
stream$putl(po, "Still here! And I got: " || int$unparse(i))
end except when empty:
stream$putl(po, "The stack is empty.")
end
end start_up</syntaxhighlight>
{{out}}
<pre>10
9
8
7
6
5
4
3
2
1
The stack is empty.</pre>
 
=={{header|COBOL}}==
{{works with|COBOL|2002}}
{{works with|OpenCOBOL|1.1}}
 
Based loosely on the C stack implementation in Evangel Quiwa's Data Structures.
 
This example (ab)uses the COPY procedure to ensure that there is a consistently-defined stack type, node type, node information type, p(redicate) type, and set of stack-utilities.
 
stack.cbl
<syntaxhighlight lang="cobol"> 01 stack.
05 head USAGE IS POINTER VALUE NULL.
</syntaxhighlight>
 
node.cbl
<syntaxhighlight lang="cobol"> 01 node BASED.
COPY node-info REPLACING
01 BY 05
node-info BY info.
05 link USAGE IS POINTER VALUE NULL.
</syntaxhighlight>
 
node-info.cbl
<syntaxhighlight lang="cobol"> 01 node-info PICTURE X(10) VALUE SPACES.
</syntaxhighlight>
 
p.cbl
<syntaxhighlight lang="cobol"> 01 p PICTURE 9.
88 nil VALUE ZERO WHEN SET TO FALSE IS 1.
88 t VALUE 1 WHEN SET TO FALSE IS ZERO.
</syntaxhighlight>
 
stack-utilities.cbl
<syntaxhighlight lang="cobol"> IDENTIFICATION DIVISION.
PROGRAM-ID. push.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY p.
COPY node.
LINKAGE SECTION.
COPY stack.
01 node-info-any PICTURE X ANY LENGTH.
PROCEDURE DIVISION USING stack node-info-any.
ALLOCATE node
CALL "pointerp" USING
BY REFERENCE ADDRESS OF node
BY REFERENCE p
END-CALL
IF nil
CALL "stack-overflow-error" END-CALL
ELSE
MOVE node-info-any TO info OF node
SET link OF node TO head OF stack
SET head OF stack TO ADDRESS OF node
END-IF
GOBACK.
END PROGRAM push.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. pop.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY p.
COPY node.
LINKAGE SECTION.
COPY stack.
COPY node-info.
PROCEDURE DIVISION USING stack node-info.
CALL "empty" USING
BY REFERENCE stack
BY REFERENCE p
END-CALL
IF t
CALL "stack-underflow-error" END-CALL
ELSE
SET ADDRESS OF node TO head OF stack
SET head OF stack TO link OF node
MOVE info OF node TO node-info
END-IF
FREE ADDRESS OF node
GOBACK.
END PROGRAM pop.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. empty.
DATA DIVISION.
LOCAL-STORAGE SECTION.
LINKAGE SECTION.
COPY stack.
COPY p.
PROCEDURE DIVISION USING stack p.
CALL "pointerp" USING
BY CONTENT head OF stack
BY REFERENCE p
END-CALL
IF t
SET t TO FALSE
ELSE
SET t TO TRUE
END-IF
GOBACK.
END PROGRAM empty.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. head.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY p.
COPY node.
LINKAGE SECTION.
COPY stack.
COPY node-info.
PROCEDURE DIVISION USING stack node-info.
CALL "empty" USING
BY REFERENCE stack
BY REFERENCE p
END-CALL
IF t
CALL "stack-underflow-error" END-CALL
ELSE
SET ADDRESS OF node TO head OF stack
MOVE info OF node TO node-info
END-IF
GOBACK.
END PROGRAM head.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. peek.
DATA DIVISION.
LOCAL-STORAGE SECTION.
LINKAGE SECTION.
COPY stack.
COPY node-info.
PROCEDURE DIVISION USING stack node-info.
CALL "head" USING
BY CONTENT stack
BY REFERENCE node-info
END-CALL
GOBACK.
END PROGRAM peek.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. pointerp.
DATA DIVISION.
LINKAGE SECTION.
01 test-pointer USAGE IS POINTER.
COPY p.
PROCEDURE DIVISION USING test-pointer p.
IF test-pointer EQUAL NULL
SET nil TO TRUE
ELSE
SET t TO TRUE
END-IF
GOBACK.
END PROGRAM pointerp.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. stack-overflow-error.
PROCEDURE DIVISION.
DISPLAY "stack-overflow-error" END-DISPLAY
STOP RUN.
END PROGRAM stack-overflow-error.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. stack-underflow-error.
PROCEDURE DIVISION.
DISPLAY "stack-underflow-error" END-DISPLAY
STOP RUN.
END PROGRAM stack-underflow-error.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. copy-stack.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY p.
COPY node-info.
LINKAGE SECTION.
COPY stack.
COPY stack REPLACING stack BY new-stack.
PROCEDURE DIVISION USING stack new-stack.
CALL "empty" USING
BY REFERENCE stack
BY REFERENCE p
END-CALL
IF nil
CALL "pop" USING
BY REFERENCE stack
BY REFERENCE node-info
END-CALL
CALL "copy-stack" USING
BY REFERENCE stack
BY REFERENCE new-stack
END-CALL
CALL "push" USING
BY REFERENCE stack
BY REFERENCE node-info
END-CALL
CALL "push" USING
BY REFERENCE new-stack
BY REFERENCE node-info
END-CALL
END-IF
GOBACK.
END PROGRAM copy-stack.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. reverse-stack.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY p.
COPY node-info.
LINKAGE SECTION.
COPY stack.
COPY stack REPLACING stack BY new-stack.
PROCEDURE DIVISION USING stack new-stack.
CALL "empty" USING
BY REFERENCE stack
BY REFERENCE p
END-CALL
IF nil
CALL "pop" USING
BY REFERENCE stack
BY REFERENCE node-info
END-CALL
CALL "push" USING
BY REFERENCE new-stack
BY REFERENCE node-info
END-CALL
CALL "reverse-stack" USING
BY REFERENCE stack
BY REFERENCE new-stack
END-CALL
CALL "push" USING
BY REFERENCE stack
BY REFERENCE node-info
END-CALL
END-IF
GOBACK.
END PROGRAM reverse-stack.
 
IDENTIFICATION DIVISION.
PROGRAM-ID. traverse-stack.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY p.
COPY node-info.
COPY stack REPLACING stack BY new-stack.
LINKAGE SECTION.
COPY stack.
PROCEDURE DIVISION USING stack.
CALL "copy-stack" USING
BY REFERENCE stack
BY REFERENCE new-stack
END-CALL
CALL "empty" USING
BY REFERENCE new-stack
BY REFERENCE p
END-CALL
IF nil
CALL "head" USING
BY CONTENT new-stack
BY REFERENCE node-info
END-CALL
DISPLAY node-info END-DISPLAY
CALL "peek" USING
BY CONTENT new-stack
BY REFERENCE node-info
END-CALL
DISPLAY node-info END-DISPLAY
CALL "pop" USING
BY REFERENCE new-stack
BY REFERENCE node-info
END-CALL
DISPLAY node-info END-DISPLAY
CALL "traverse-stack" USING
BY REFERENCE new-stack
END-CALL
END-IF
GOBACK.
END PROGRAM traverse-stack.
</syntaxhighlight>
 
stack-test.cbl
<syntaxhighlight lang="cobol"> IDENTIFICATION DIVISION.
PROGRAM-ID. stack-test.
DATA DIVISION.
LOCAL-STORAGE SECTION.
COPY stack.
COPY stack REPLACING stack BY new-stack.
PROCEDURE DIVISION.
CALL "push" USING
BY REFERENCE stack
BY CONTENT "daleth"
END-CALL
CALL "push" USING
BY REFERENCE stack
BY CONTENT "gimel"
END-CALL
CALL "push" USING
BY REFERENCE stack
BY CONTENT "beth"
END-CALL
CALL "push" USING
BY REFERENCE stack
BY CONTENT "aleph"
END-CALL
CALL "traverse-stack" USING
BY REFERENCE stack
END-CALL
CALL "reverse-stack" USING
BY REFERENCE stack
BY REFERENCE new-stack
END-CALL
CALL "traverse-stack" USING
BY REFERENCE new-stack
END-CALL
STOP RUN.
END PROGRAM stack-test.
 
COPY stack-utilities.
</syntaxhighlight>
 
{{out}}
<pre>
aleph
aleph
beth
beth
beth
gimel
gimel
gimel
daleth
daleth
daleth
daleth
daleth
daleth
gimel
gimel
gimel
beth
beth
beth
aleph
aleph
aleph
</pre>
 
=={{header|CoffeeScript}}==
<syntaxhighlight lang="coffeescript">stack = []
stack.push 1
stack.push 2
console.log stack
console.log stack.pop()
console.log stack</syntaxhighlight>
 
=={{header|Common Lisp}}==
It's a bit unusual to write a wrapper for a stack in Common Lisp; built-in cons-based lists work just fine. Nonetheless, here's an implementation where the list is wrapped in a structure, providing a stateful solution.
<syntaxhighlight lang="lisp">(defstruct stack
elements)
 
Line 373 ⟶ 2,229:
(push element (stack-elements stack)))
 
(defun stack-pop (stack)(deftype Stack [elements])
(pop (stack-elements stack)))
 
(defun stack-empty (stack)
Line 383 ⟶ 2,238:
 
(defun stack-peek (stack)
(stack-top stack))</langsyntaxhighlight>
 
=={{header|Component Pascal}}==
Works with BlackBox Component Builder
<syntaxhighlight lang="oberon2">
MODULE Stacks;
IMPORT StdLog;
 
TYPE
(* some pointers to records *)
Object* = POINTER TO ABSTRACT RECORD END;
Integer = POINTER TO RECORD (Object)
i: INTEGER
END;
Point = POINTER TO RECORD (Object)
x,y: REAL
END;
 
Node = POINTER TO LIMITED RECORD
next- : Node;
data-: ANYPTR;
END;
 
(* Stack *)
Stack* = POINTER TO RECORD
top- : Node;
END;
PROCEDURE (dn: Object) Show*, NEW, ABSTRACT;
PROCEDURE (i: Integer) Show*;
BEGIN
StdLog.String("Integer(");StdLog.Int(i.i);StdLog.String(");");StdLog.Ln
END Show;
PROCEDURE (p: Point) Show*;
BEGIN
StdLog.String("Point(");StdLog.Real(p.x);StdLog.Char(',');
StdLog.Real(p.y);StdLog.String(");");StdLog.Ln
END Show;
PROCEDURE (s: Stack) Init, NEW;
BEGIN
s.top := NIL;
END Init;
PROCEDURE (s: Stack) Push*(data: ANYPTR), NEW;
VAR
n: Node;
BEGIN
NEW(n);n.next := NIL;n.data := data;
IF s.top = NIL THEN
s.top := n
ELSE
n.next := s.top;
s.top := n
END
END Push;
PROCEDURE (s: Stack) Pop*(): ANYPTR, NEW;
VAR
x: ANYPTR;
BEGIN
IF s.top # NIL THEN
x := s.top.data;
s.top := s.top.next
ELSE
x := NIL
END;
RETURN x
END Pop;
PROCEDURE (s: Stack) Empty*(): BOOLEAN, NEW;
BEGIN
RETURN s.top = NIL
END Empty;
PROCEDURE NewStack*(): Stack;
VAR
s: Stack;
BEGIN
NEW(s);s.Init;
RETURN s
END NewStack;
PROCEDURE NewInteger*(data: INTEGER): Integer;
VAR
i: Integer;
BEGIN
NEW(i);i.i := data;
RETURN i
END NewInteger;
PROCEDURE NewPoint*(x,y: REAL): Point;
VAR
p: Point;
BEGIN
NEW(p);p.x := x;p.y := y;
RETURN p
END NewPoint;
PROCEDURE TestStack*;
VAR
s: Stack;
BEGIN
s := NewStack();
s.Push(NewInteger(1));
s.Push(NewPoint(2.0,3.4));
s.Pop()(Object).Show();
s.Pop()(Object).Show();
END TestStack;
END Stacks.
</syntaxhighlight>
 
Execute: ^Q Stacks.TestStack<br/>
{{out}}
<pre>
Point( 2.0, 3.4);
Integer( 1);
</pre>
 
=={{header|Crystal}}==
 
<syntaxhighlight lang="ruby">stack = [] of Int32
(1..10).each do |x|
stack.push x
end
 
10.times do
puts stack.pop
end</syntaxhighlight>
 
'''Output:'''
 
<pre>
10
9
8
7
6
5
4
3
2
1
</pre>
 
=={{header|D}}==
Implemented aGeneric stack class byimplemented usingwith a sequencedynamic array.
<syntaxhighlight lang="d">import std.array;
<lang d>
 
module stack ;
class Stack(T) {
private T[] content = null items;
 
void push(T top) { content ~= top ; }
@property bool empty() { return items.empty(); }
T pop() {
 
if (empty)
void push(T top) { items ~= top; }
throw new Exception("Stack Empty") ;
 
T top = content[$-1] ;
T pop() {
content.length = content.length - 1 ;
if (this.empty)
return top ;
throw new Exception("Empty Stack.");
}
auto top = items.back;
bool empty() { return content.length == 0 ; }
items.popBack();
return top;
}
}
</lang>
=={{header|E}}==
 
void main() {
The standard FlexList data structure provides operations for use as a stack.
auto s = new Stack!int();
s.push(10);
s.push(20);
assert(s.pop() == 20);
assert(s.pop() == 10);
assert(s.empty());
}</syntaxhighlight>
 
=={{header|Delphi}}==
? def l := [].diverge()
<syntaxhighlight lang="delphi">program Stack;
# value: [].diverge()
 
{$APPTYPE CONSOLE}
 
uses Generics.Collections;
 
var
lStack: TStack<Integer>;
begin
lStack := TStack<Integer>.Create;
try
lStack.Push(1);
lStack.Push(2);
lStack.Push(3);
Assert(lStack.Peek = 3); // 3 should be at the top of the stack
 
Writeln(lStack.Pop); // 3
Writeln(lStack.Pop); // 2
Writeln(lStack.Pop); // 1
Assert(lStack.Count = 0); // should be empty
finally
lStack.Free;
end;
end.</syntaxhighlight>
 
=={{header|DWScript}}==
Dynamic arrays have pseudo-methods that allow to treat them as a stack.
<syntaxhighlight lang="delphi">
var stack: array of Integer;
 
stack.Push(1);
stack.Push(2);
stack.Push(3);
 
PrintLn(stack.Pop); // 3
PrintLn(stack.Pop); // 2
PrintLn(stack.Pop); // 1
 
Assert(stack.Length = 0); // assert empty
</syntaxhighlight>
 
=={{header|Dyalect}}==
 
{{trans|Swift}}
 
<syntaxhighlight lang="dyalect">type Stack() {
var xs = []
}
func Stack.IsEmpty() => this!xs.Length() == 0
? l.push(1)
? l.push(2)
? l
# value: [1, 2].diverge()
func Stack.Peek() => this!xs[this!xs.Length() - 1]
? l.pop()
# value: 2
func Stack.Pop() {
? l.size().aboveZero()
var e = this!xs[this!xs.Length() - 1]
# value: true
this!xs.RemoveAt(this!xs.Length() - 1)
return e
}
func Stack.Push(item) => this!xs.Add(item)
? l.last()
# value: 1
var stack = Stack()
? l.pop()
stack.Push(1)
# value: 1
stack.Push(2)
print(stack.Pop())
? l.size().aboveZero()
print(stack.Peek())
# value: false
stack.Pop()
print(stack.IsEmpty())</syntaxhighlight>
 
{{out}}
 
<pre>2
1
true</pre>
 
=={{header|Déjà Vu}}==
<syntaxhighlight lang="dejavu">local :stack [] #lists used to be stacks in DV
 
push-to stack 1
push-to stack 2
push-to stack 3
 
!. pop-from stack #prints 3
!. pop-from stack #prints 2
!. pop-from stack #prints 1
 
if stack: #empty lists are falsy
error #this stack should be empty now!</syntaxhighlight>
 
=={{header|Diego}}==
Diego has a <code>stack</code> object and posit:
<syntaxhighlight lang="diego">set_ns(rosettacode)_me();
 
add_stack({int},a)_values(1..4); // 1,2,3,4 (1 is first/bottom, 4 is last/top)
with_stack(a)_pop(); // 1,2,3
with_stack(a)_push()_v(5,6); // 1,2,3,5,6
 
add_var({int},b)_value(7);
with_stack(a)_push[b]; // 1,2,3,5,6,7
 
with_stack(a)_pluck()_at(2); // callee will return `with_stack(a)_err(pluck invalid with stack);`
 
me_msg()_stack(a)_top(); // "7"
me_msg()_stack(a)_last(); // "7"
me_msg()_stack(a)_peek(); // "7"
 
me_msg()_stack(a)_bottom(); // "1"
me_msg()_stack(a)_first(); // "1"
me_msg()_stack(a)_peer(); // "1"
 
me_msg()_stack(a)_isempty(); // "false"
with_stack(a)_empty();
with_stack(a)_msg()_isempty()_me(); // "true" (alternative syntax)
 
me_msg()_stack(a)_history()_all(); // returns th entire history of stack 'a' since its creation
 
reset_ns[];</syntaxhighlight>
<code>stack</code> is a derivative of <code>array</code>, so arrays can also be used as stacks.
 
=={{header|E}}==
The standard FlexList data structure provides operations for use as a stack.
<syntaxhighlight lang="e">? def l := [].diverge()
# value: [].diverge()
 
? l.push(1)
? l.push(2)
? l
# value: [1, 2].diverge()
 
? l.pop()
# value: 2
 
? l.size().aboveZero()
# value: true
 
? l.last()
# value: 1
 
? l.pop()
# value: 1
 
? l.size().aboveZero()
# value: false</syntaxhighlight>
Here's a stack implemented out of a reference to a linked list:
<syntaxhighlight lang="e">def makeStack() {
var store := null
def stack {
to push(x) { store := [x, store] }
to pop() { def [x, next] := store; store := next; return x }
to last() { return store[0] }
to empty() { return (store == null) }
}
return stack
}
 
? def s := makeStack() {
# value: <stack>
var store := null
def stack {
to push(x) { store := [x, store] }
to pop() { def [x, next] := store; store := next; return x }
to last() { return store[0] }
to empty() { return (store == null) }
}
return stack
}
 
? def s := makeStack.push(1)
? s.push(2)
# value: <stack>
? s.last()
# value: 2
? s.push(1)
 
? s.push(2)
? s.lastpop()
# value: 2
 
? s.empty()
# value: false
 
? s.pop()
# value: 1
 
? s.empty()
# value: true</syntaxhighlight>
 
=={{header|EasyLang}}==
<syntaxhighlight>
stack[] = [ ]
proc push v . .
stack[] &= v
.
func pop .
lng = len stack[]
if lng = 0
return 0
.
r = stack[lng]
len stack[] -1
return r
.
func empty .
return if len stack[] = 0
.
push 2
push 11
push 34
while empty = 0
print pop
.
</syntaxhighlight>
 
=={{header|EchoLisp}}==
Named stacks are native objects. The following demonstrates the available operations :
<syntaxhighlight lang="lisp">
; build stack [0 1 ... 9 (top)] from a list
(list->stack (iota 10) 'my-stack)
(stack-top 'my-stack) → 9
(pop 'my-stack) → 9
(stack-top 'my-stack) → 8
(push 'my-stack '🐸) ; any kind of lisp object in the stack
(stack-empty? 'my-stack) → #f
(stack->list 'my-stack) ; convert stack to list
→ (0 1 2 3 4 5 6 7 8 🐸)
(stack-swap 'my-stack) ; swaps two last items
→ 8 ; new top
(stack->list 'my-stack)
→ (0 1 2 3 4 5 6 7 🐸 8) ; swapped
(while (not (stack-empty? 'my-stack)) (pop 'my-stack)) ; pop until empty
(stack-empty? 'my-stack) → #t ; true
 
(push 'my-stack 7)
my-stack ; a stack is not a variable, nor a symbol - cannot be evaluated
⛔ error: #|user| : unbound variable : my-stack
(stack-top 'my-stack) → 7
</syntaxhighlight>
 
=={{header|Eiffel}}==
<syntaxhighlight lang="eiffel">
class
STACK_ON_ARRAY
 
create
make
 
feature -- Implementation
 
empty: BOOLEAN
do
Result := stack.is_empty
ensure
empty: Result = (stack.count = 0)
end
 
push (item: ANY)
do
stack.force (item, stack.count)
ensure
pushed: stack [stack.upper] = item
growth: stack.count = old stack.count + 1
end
 
pop: ANY
require
not_empty: not empty
do
Result := stack.at (stack.upper)
stack.remove_tail (1)
ensure
reduction: stack.count = old stack.count - 1
end
 
feature {NONE} -- Initialization
 
stack: ARRAY [ANY]
 
make
do
create stack.make_empty
end
 
end
</syntaxhighlight>
 
=={{header|Elena}}==
<syntaxhighlight lang="elena">public program()
{
var stack := new system'collections'Stack();
stack.push(2);
? s.pop()
# value: 2
var isEmpty := stack.Length == 0;
? s.empty()
# value: false
var item := stack.peek(); // Peek without Popping.
? s.pop()
# value: 1
item := stack.pop()
? s.empty()
}</syntaxhighlight>
# value: true
 
=={{header|Elisa}}==
This is a generic Stack component based on arrays. See how in Elisa [http://jklunder.home.xs4all.nl/elisa/part01/doc120.html generic components] are defined.
<syntaxhighlight lang="elisa"> component GenericStack ( Stack, Element );
type Stack;
Stack (MaxSize = integer) -> Stack;
Empty ( Stack ) -> boolean;
Full ( Stack ) -> boolean;
Push ( Stack, Element) -> nothing;
Pull ( Stack ) -> Element;
begin
Stack(MaxSize) =
Stack:[ MaxSize; index:=0; area=array (Element, MaxSize) ];
Empty( stack ) = (stack.index <= 0);
Full ( stack ) = (stack.index >= stack.MaxSize);
Push ( stack, element ) =
[ exception (Full (stack), "Stack Overflow");
stack.index:=stack.index + 1;
stack.area[stack.index]:=element ];
Pull ( stack ) =
[ exception (Empty (stack), "Stack Underflow");
stack.index:=stack.index - 1;
stack.area[stack.index + 1] ];
end component GenericStack;</syntaxhighlight>
Another example of a generic Stack component is based on an unlimited sequence. A sequence is a uni-directional list. See how Elisa defines [http://jklunder.home.xs4all.nl/elisa/part02/doc010.html sequences]. The component has the same interface as the array based version.
<syntaxhighlight lang="elisa">component GenericStack ( Stack, ElementType );
type Stack;
Stack(MaxSize = integer) -> Stack;
Empty( Stack ) -> boolean;
Full ( Stack ) -> boolean;
Push ( Stack, ElementType)-> nothing;
Pull ( Stack ) -> ElementType;
begin
type sequence = term;
ElementType & sequence => sequence;
nil = null (sequence);
 
head (sequence) -> ElementType;
head (X & Y) = ElementType:X;
 
tail (sequence) -> sequence;
tail (X & Y) = Y;
 
Stack (Size) = Stack:[ list = nil ];
Empty ( stack ) = (stack.list == nil);
Full ( stack ) = false;
Push ( stack, ElementType ) = [ stack.list:= ElementType & stack.list ];
Pull ( stack ) = [ exception (Empty (stack), "Stack Underflow");
Head = head(stack.list); stack.list:=tail(stack.list); Head];
end component GenericStack;</syntaxhighlight>
Both versions give the same answers to the following tests:
<syntaxhighlight lang="elisa">use GenericStack (StackofBooks, Book);
type Book = text;
BookStack = StackofBooks(50);
 
Push (BookStack, "Peter Pan");
Push (BookStack, "Alice in Wonderland");
 
Pull (BookStack)?
"Alice in Wonderland"
 
Pull (BookStack)?
"Peter Pan"
 
Pull (BookStack)?
***** Exception: Stack Underflow</syntaxhighlight>
 
=={{header|Elixir}}==
{{trans|Erlang}}
<syntaxhighlight lang="elixir">defmodule Stack do
def new, do: []
def empty?([]), do: true
def empty?(_), do: false
def pop([h|t]), do: {h,t}
def push(h,t), do: [h|t]
def top([h|_]), do: h
end</syntaxhighlight>
 
Example:
<pre>
iex(2)> stack = Stack.new
[]
iex(3)> Stack.empty?(stack)
true
iex(4)> newstack = List.foldl([1,2,3,4,5], stack, fn x,acc -> Stack.push(x,acc) end)
[5, 4, 3, 2, 1]
iex(5)> Stack.top(newstack)
5
iex(6)> {popped, poppedstack} = Stack.pop(newstack)
{5, [4, 3, 2, 1]}
iex(7)> Stack.empty?(newstack)
false
</pre>
 
=={{header|Erlang}}==
Erlang has no built-in stack, but its lists behave basically the same way. A stack module can be implemented as a simple wrapper around lists:
<syntaxhighlight lang="erlang">-module(stack).
<lang erlang>
-module(stack).
-export([empty/1, new/0, pop/1, push/2, top/1]).
 
Line 477 ⟶ 2,816:
push(H,T) -> [H|T].
 
top([H|_]) -> H.</syntaxhighlight>
</lang>
 
Note that as Erlang doesn't have mutable data structure (destructive updates), pop returns the popped element and the new stack as a tuple.
 
The module is tested this way:
<langsyntaxhighlight lang="erlang">1> c(stack).
1> c(stack).
{ok,stack}
2> Stack = stack:new().
Line 497 ⟶ 2,833:
false
7> stack:empty(stack:new()).
true</syntaxhighlight>
true
</lang>
 
=={{header|F_Sharp|F#}}==
.NET provides a mutable stack type in <code>System.Collections.Generic.Stack</code>.
 
A list-based immutable stack type could be implemented like this:
<syntaxhighlight lang="fsharp">type Stack<'a> //'//(workaround for syntax highlighting problem)
(?items) =
let items = defaultArg items []
 
member x.Push(A) = Stack(A::items)
 
member x.Pop() =
match items with
| x::xr -> (x, Stack(xr))
| [] -> failwith "Stack is empty."
 
member x.IsEmpty() = items = []
 
// example usage
let anEmptyStack = Stack<int>()
let stack2 = anEmptyStack.Push(42)
printfn "%A" (stack2.IsEmpty())
let (x, stack3) = stack2.Pop()
printfn "%d" x
printfn "%A" (stack3.IsEmpty())</syntaxhighlight>
 
=={{header|Factor}}==
Factor is a stack based language, but also provides stack "objects", because
all resizable sequences can be treated as stacks (see [http://docs.factorcode.org/content/article-sequences-stacks.html docs]). Typically, a vector is used:
<syntaxhighlight lang="factor"> V{ 1 2 3 } {
[ 6 swap push ]
[ "hi" swap push ]
[ "Vector is now: " write . ]
[ "Let's pop it: " write pop . ]
[ "Vector is now: " write . ]
[ "Top is: " write last . ] } cleave
 
Vector is now: V{ 1 2 3 6 "hi" }
Let's pop it: "hi"
Vector is now: V{ 1 2 3 6 }
Top is: 6</syntaxhighlight>
 
=={{header|Forth}}==
<syntaxhighlight lang="forth">: stack ( size -- )
create here cell+ , cells allot ;
 
: push ( n st -- ) tuck @ ! cell swap +! ;
: pop ( st -- n ) -cell over +! @ @ ;
: empty? ( st -- ? ) dup @ - cell+ 0= ;
 
10 stack st
 
1 st push
2 st push
3 st push
st empty? . \ 0 (false)
st pop . st pop . st pop . \ 3 2 1
st empty? . \ -1 (true)</syntaxhighlight>
 
=={{header|Fortran}}==
This solution can easily be adapted to data types other than floating point numbers.
<syntaxhighlight lang="fortran">module mod_stack
 
implicit none
type node
! data entry in each node
real*8, private :: data
! pointer to the next node of the linked list
type(node), pointer, private :: next
end type node
private node
 
type stack
! pointer to first element of stack.
type(node), pointer, private :: first
! size of stack
integer, private :: len=0
contains
procedure :: pop
procedure :: push
procedure :: peek
procedure :: getSize
procedure :: clearStack
procedure :: isEmpty
end type stack
 
contains
 
function pop(this) result(x)
class(stack) :: this
real*8 :: x
type(node), pointer :: tmp
if ( this%len == 0 ) then
print*, "popping from empty stack"
!stop
end if
tmp => this%first
x = this%first%data
this%first => this%first%next
deallocate(tmp)
this%len = this%len -1
end function pop
 
subroutine push(this, x)
real*8 :: x
class(stack), target :: this
type(node), pointer :: new, tmp
allocate(new)
new%data = x
if (.not. associated(this%first)) then
this%first => new
else
tmp => this%first
this%first => new
this%first%next => tmp
end if
this%len = this%len + 1
end subroutine push
 
function peek(this) result(x)
class(stack) :: this
real*8 :: x
x = this%first%data
end function peek
 
function getSize(this) result(n)
class(stack) :: this
integer :: n
n = this%len
end function getSize
 
function isEmpty(this) result(empty)
class(stack) :: this
logical :: empty
if ( this%len > 0 ) then
empty = .FALSE.
else
empty = .TRUE.
end if
end function isEmpty
 
subroutine clearStack(this)
class(stack) :: this
type(node), pointer :: tmp
integer :: i
if ( this%len == 0 ) then
return
end if
do i = 1, this%len
tmp => this%first
if ( .not. associated(tmp)) exit
this%first => this%first%next
deallocate(tmp)
end do
this%len = 0
end subroutine clearStack
end module mod_stack
 
program main
use mod_stack
type(stack) :: my_stack
integer :: i
real*8 :: dat
do i = 1, 5, 1
dat = 1.0 * i
call my_stack%push(dat)
end do
do while ( .not. my_stack%isEmpty() )
print*, my_stack%pop()
end do
call my_stack%clearStack()
end program main</syntaxhighlight>
 
=={{header|Free Pascal}}==
===Delphi adaptation===
Example taken and adapted from the Delphi entry.
<syntaxhighlight lang="pascal">program Stack;
{$IFDEF FPC}{$MODE DELPHI}{$IFDEF WINDOWS}{$APPTYPE CONSOLE}{$ENDIF}{$ENDIF}
{$ASSERTIONS ON}
uses Generics.Collections;
var
: push ( n st -- ) tuck @ ! cell swap +! ;
lStack: TStack<Integer>;
: pop ( st -- n ) -cell over +! @ @ ;
begin
: empty? ( st -- ? ) dup @ - cell+ 0= ;
lStack := TStack<Integer>.Create;
try
lStack.Push(1);
lStack.Push(2);
lStack.Push(3);
Assert(lStack.Peek = 3); // 3 should be at the top of the stack
Write(lStack.Pop:2); // 3
Write(lStack.Pop:2); // 2
Writeln(lStack.Pop:2); // 1
Assert(lStack.Count = 0, 'Stack is not empty'); // should be empty
finally
lStack.Free;
end;
end.</syntaxhighlight>
<pre>
Output:
3 2 1
</pre>
 
===Object version from scratch===
{{works with|Free Pascal|version 3.2.0 }}
<syntaxhighlight lang="pascal">
PROGRAM StackObject.pas;
{$IFDEF FPC}
{$mode objfpc}{$H+}{$J-}{$m+}{$R+}
{$ELSE}
{$APPTYPE CONSOLE}
{$ENDIF}
(*)
 
Free Pascal Compiler version 3.2.0 [2020/06/14] for x86_64
TheStack free and readable alternative at C/C++ Sidxeeds
compiles natively to almost any platform, including raSidxberry PI *
Can run independently from DELPHI / Lazarus
 
For debian Linux: apt -y install fpc
It contains a text IDE called fp
 
This is an experiment for a stack that can handle almost any
simple type of variable.
 
What happens after retrieving the variable is TBD by you.
 
https://www.freepascal.org/advantage.var
(*)
 
 
USES
Classes ,
Crt ,
Variants ;
{$WARN 6058 off : Call to subroutine "$1" marked as inline is not inlined} // Use for variants
 
TYPE
 
Stack = OBJECT
CONST
CrLf = #13#10 ;
TYPE
 
VariantArr = array of variant ;
PRIVATE
 
Ar : VariantArr ;
 
{$MACRO ON}
{$DEFINE STACKSIZE := Length ( Ar ) * Ord ( Length ( Ar ) > 0 ) }
{$DEFINE TOP := STACKSIZE - 1 * Ord ( STACKSIZE > 0 ) }
{$DEFINE SLEN := length ( Ar [ TOP ] ) * Ord ( Length ( Ar [ TOP ] ) > 0 ) }
 
FUNCTION IsEmpty : boolean ;
PROCEDURE Print ;
FUNCTION Pop : variant ;
FUNCTION Peep : variant ;
PROCEDURE Push ( item : variant ) ;
FUNCTION SecPop : variant ;
 
PUBLIC
CONSTRUCTOR Create ;
END;
 
CONSTRUCTOR Stack.Create ;
 
BEGIN
SetLength ( Ar, STACKSIZE ) ;
END;
 
FUNCTION Stack.IsEmpty : boolean ;
BEGIN
IsEmpty := ( STACKSIZE < 1 ) ;
END;
 
 
PROCEDURE Stack.Print ;
 
VAR
i : shortint ;
BEGIN
IF ( TOP < 1 ) or ( IsEmpty ) THEN
BEGIN
WriteLn ( CrLf + '<empty stack>' ) ;
EXIT ;
END;
WriteLn ( CrLf , '<top>') ;
FOR i := ( TOP ) DOWNTO 0 DO WriteLn ( Ar [ i ] ) ;
WriteLn ( '<bottom>' ) ;
END;
 
 
FUNCTION Stack.Pop : variant ;
 
BEGIN
IF IsEmpty THEN EXIT ;
Pop := Ar [ TOP ] ;
SetLength ( Ar, TOP ) ;
END;
 
 
FUNCTION Stack.Peep : variant ;
 
BEGIN
IF IsEmpty THEN EXIT ;
Peep := Ar [ TOP ] ;
END;
 
 
PROCEDURE Stack.Push ( item : variant ) ;
BEGIN
SetLength ( Ar, STACKSIZE + 1 ) ;
Ar [ TOP ] := item ;
END;
 
 
FUNCTION Stack.SecPop : variant ;
(*) Pop and Wipe (*)
BEGIN
IF IsEmpty THEN EXIT ;
SecPop := Ar [ TOP ] ;
Ar [ TOP ] := StringOfChar ( #255 , SLEN ) ;
Ar [ TOP ] := StringOfChar ( #0 , SLEN ) ;
SetLength ( Ar, TOP ) ;
END;
 
VAR
n : integer ;
r : real ;
S : string ;
So : Stack ;
 
 
BEGIN
 
So.Create ;
So.Print ;
n := 23 ;
So.Push ( n ) ;
S := '3 guesses ' ;
So.Push ( S ) ;
r := 1.23 ;
So.Push ( r ) ;
WriteLn ( 'Peep : ', So.Peep ) ;
So.Push ( 'Nice Try' ) ;
So.Print ;
WriteLn ;
WriteLn ( 'SecPop : ',So.SecPop ) ;
WriteLn ( 'SecPop : ',So.SecPop ) ;
WriteLn ( 'SecPop : ',So.SecPop ) ;
WriteLn ( 'SecPop : ',So.SecPop ) ;
So.Print ;
END.
 
</syntaxhighlight>JPD 2021/07/03
 
Output:
 
<empty stack>
 
Peep : 1.23
 
<top>
 
Nice Try
 
1.23
 
3 guesses
 
23
 
<bottom>
 
SecPop : Nice Try
 
SecPop : 1.23
 
SecPop : 3 guesses
 
SecPop : 23
 
<empty stack>
 
=={{header|FreeBASIC}}==
We first use a macro to define a generic Stack type :
<syntaxhighlight lang="freebasic">' FB 1.05.0 Win64
 
' stack_rosetta.bi
' simple generic Stack type
 
#Define Stack(T) Stack_##T
 
#Macro Declare_Stack(T)
Type Stack(T)
Public:
Declare Constructor()
Declare Destructor()
Declare Property capacity As Integer
Declare Property count As Integer
Declare Property empty As Boolean
Declare Property top As T
Declare Function pop() As T
Declare Sub push(item As T)
Private:
a(any) As T
count_ As Integer = 0
Declare Function resize(size As Integer) As Integer
End Type
 
Constructor Stack(T)()
Redim a(0 To 0) '' create a default T instance for various purposes
End Constructor
 
Destructor Stack(T)()
Erase a
End Destructor
 
Property Stack(T).capacity As Integer
10 stack st
Return UBound(a)
End Property
Property Stack(T).count As Integer
1 st push
Return count_
2 st push
End Property
3 st push
 
st empty? . \ 0 (false)
Property Stack(T).empty As Boolean
st pop . st pop . st pop . \ 3 2 1
Return count_ = 0
st empty? . \ -1 (true)
End Property
 
Property Stack(T).top As T
If count_ > 0 Then
Return a(count_)
End If
Print "Error: Attempted to access 'top' element of an empty stack"
Return a(0) '' return default element
End Property
 
Function Stack(T).pop() As T
If count_ > 0 Then
Dim value As T = a(count_)
a(count_) = a(0) '' zero element to be removed
count_ -= 1
Return value
End If
Print "Error: Attempted to remove 'top' element of an empty stack"
Return a(0) '' return default element
End Function
 
Sub Stack(T).push(item As T)
Dim size As Integer = UBound(a)
count_ += 1
If count_ > size Then
size = resize(size)
Redim Preserve a(0 to size)
End If
a(count_) = item
End Sub
 
Function Stack(T).resize(size As Integer) As Integer
If size = 0 Then
size = 4
ElseIf size <= 32 Then
size = 2 * size
Else
size += 32
End If
Return size
End Function
#EndMacro</syntaxhighlight>
 
We now use this type to create a Stack of Dog instances :
<syntaxhighlight lang="freebasic">' FB 1.05.0 Win64
 
#Include "stack_rosetta.bi"
 
Type Dog
name As String
age As Integer
Declare Constructor
Declare Constructor(name_ As string, age_ As integer)
Declare Operator Cast() As String
end type
 
Constructor Dog '' default constructor
End Constructor
 
Constructor Dog(name_ As String, age_ As Integer)
name = name_
age = age_
End Constructor
 
Operator Dog.Cast() As String
Return "[" + name + ", " + Str(age) + "]"
End Operator
 
Declare_Stack(Dog) '' expand Stack type for Dog instances
 
Dim dogStack As Stack(Dog)
 
Var cerberus = Dog("Cerberus", 10)
Var rover = Dog("Rover", 3)
Var puppy = Dog("Puppy", 0)
With dogStack '' push these Dog instances onto the stack
.push(cerberus)
.push(rover)
.push(puppy)
End With
Print "Number of dogs on the stack :" ; dogStack.count
Print "Capacity of dog stack :" ; dogStack.capacity
Print "Top dog : "; dogStack.top
dogStack.pop()
Print "Top dog now : "; dogStack.top
Print "Number of dogs on the stack :" ; dogStack.count
dogStack.pop()
Print "Top dog now : "; dogStack.top
Print "Number of dogs on the stack :" ; dogStack.count
Print "Is stack empty now : "; dogStack.empty
Print
Print "Press any key to quit"
Sleep</syntaxhighlight>
 
{{out}}
<pre>
Number of dogs on the stack : 3
Capacity of dog stack : 4
Top dog : [Puppy, 0]
Top dog now : [Rover, 3]
Number of dogs on the stack : 2
Top dog now : [Cerberus, 10]
Number of dogs on the stack : 1
Is stack empty now : false
</pre>
 
=={{header|Frink}}==
Frink's <CODE>array</CODE> class has all of the methods to make it usable as a stack or a deque. The methods are called <CODE><I>array</I>.push[<I>x</I>]</CODE>, <CODE><I>array</I>.pop[]</CODE>, and <CODE><I>array</I>.isEmpty[]</CODE>
<syntaxhighlight lang="frink">a = new array
a.push[1]
a.push[2]
a.peek[]
while ! a.isEmpty[]
println[a.pop[]]</syntaxhighlight>
 
=={{header|Genie}}==
<syntaxhighlight lang="genie">[indent=4]
/*
Stack, in Genie, with GLib double ended Queues
valac stack.gs
*/
init
var stack = new Queue of int()
 
// push
stack.push_tail(2)
stack.push_tail(1)
 
// pop (and peek at top)
print stack.pop_tail().to_string()
print stack.peek_tail().to_string()
 
// empty
print "stack size before clear: " + stack.get_length().to_string()
stack.clear()
print "After clear, stack.is_empty(): " + stack.is_empty().to_string()</syntaxhighlight>
 
{{out}}
<pre>prompt$ valac stack.gs
prompt$ ./stack
1
2
stack size before clear: 1
After clear, stack.is_empty(): true</pre>
 
=={{header|Go}}==
Go slices make excellent stacks without defining any extra types, functions, or methods. For example, to keep a stack of integers, simply declare one as,
<syntaxhighlight lang="go">var intStack []int</syntaxhighlight>
Use the built in append function to push numbers on the stack:
<syntaxhighlight lang="go">intStack = append(intStack, 7)</syntaxhighlight>
Use a slice expression with the built in len function to pop from the stack:
<syntaxhighlight lang="go">popped, intStack = intStack[len(intStack)-1], intStack[:len(intStack)-1]</syntaxhighlight>
The test for an empty stack:
<syntaxhighlight lang="go">len(intStack) == 0</syntaxhighlight>
And to peek at the top of the stack:
<syntaxhighlight lang="go">intStack[len(intStack)-1]</syntaxhighlight>
It is idiomatic Go to use primitive language features where they are sufficient, and define helper functions or types and methods only as they make sense for a particular situation. Below is an example using a type with methods and idiomatic "ok" return values to avoid panics. It is only an example of something that might make sense in some situation.
<syntaxhighlight lang="go">package main
 
import "fmt"
 
type stack []interface{}
 
func (k *stack) push(s interface{}) {
*k = append(*k, s)
}
 
func (k *stack) pop() (s interface{}, ok bool) {
if k.empty() {
return
}
last := len(*k) - 1
s = (*k)[last]
*k = (*k)[:last]
return s, true
}
 
func (k *stack) peek() (s interface{}, ok bool) {
if k.empty() {
return
}
last := len(*k) - 1
s = (*k)[last]
return s, true
}
 
func (k *stack) empty() bool {
return len(*k) == 0
}
 
func main() {
var s stack
fmt.Println("new stack:", s)
fmt.Println("empty?", s.empty())
s.push(3)
fmt.Println("push 3. stack:", s)
fmt.Println("empty?", s.empty())
s.push("four")
fmt.Println(`push "four" stack:`, s)
if top, ok := s.peek(); ok {
fmt.Println("top value:", top)
} else {
fmt.Println("nothing on stack")
}
if popped, ok := s.pop(); ok {
fmt.Println(popped, "popped. stack:", s)
} else {
fmt.Println("nothing to pop")
}
}</syntaxhighlight>
{{out}}
<pre>
new stack: []
empty? true
push 3. stack: [3]
empty? false
push "four" stack: [3 four]
top value: four
four popped. stack: [3]
</pre>
 
=={{header|GDScript}}==
In GDScript there is built-in Array class, that implements either 'push', 'pop', 'top' and 'empty' methods. Method names are:
<ul>
<li>push -> push_back</li>
<li>pop -> pop_back</li>
<li>top -> back</li>
<li>empty -> is_empty</li>
</ul>
<syntaxhighlight lang="gdscript">
extends Node2D
 
func _ready() -> void:
# Empty stack creation.
var stack : Array = []
# In Godot 4.2.1 nothing happens here.
stack.pop_back()
if stack.is_empty():
print("Stack is empty.")
stack.push_back(3)
stack.push_back("Value")
stack.push_back(1.5e32)
print(stack)
print("Last element is: " + str(stack.back()))
stack.pop_back()
print(stack)
print("Last element is: " + str(stack.back()))
if not stack.is_empty():
print("Stack is not empty.")
return
</syntaxhighlight>
{{out}}
<pre>
Stack is empty.
[3, "Value", 149999999999999999042044051849216]
Last element is: 149999999999999999042044051849216
[3, "Value"]
Last element is: Value
Stack is not empty.
</pre>
 
=={{header|Groovy}}==
Line 523 ⟶ 3,547:
Of course, these stack semantics are not ''exclusive''. Elements of the list can still be accessed and manipulated in myriads of other ways.
 
If you need exclusive stack semantics, you can use the <code>java.util.Stack</code> class, as demonstrated in the [[Stack#Java|Java]] example.
<lang groovy>def stack = []
 
<syntaxhighlight lang="groovy">def stack = []
assert stack.empty
 
Line 559 ⟶ 3,585:
assert stack.empty
 
try { stack.pop() } catch (NoSuchElementException e) { println e.message }</langsyntaxhighlight>
 
{{out}}
Output:
<pre>[55, 21, kittens]
[55, 21]
Line 571 ⟶ 3,597:
 
=={{header|Haskell}}==
 
The Haskell solution is trivial, using a list. Note that <code>pop</code> returns both the element and the changed stack, to remain purely functional.
<syntaxhighlight lang="haskell">type Stack a = [a]
 
<lang haskell>type Stack a = [a]
 
create :: Stack a
Line 591 ⟶ 3,615:
peek :: Stack a -> a
peek [] = error "Stack empty"
peek (x:_) = x</langsyntaxhighlight>
 
We can make a stack that can be destructively popped by hiding the list inside a <code>State</code> monad.
<syntaxhighlight lang="haskell">import Control.Monad.State
 
<lang haskell>import Control.Monad.State
 
type Stack a b = State [a] b
Line 616 ⟶ 3,638:
 
nonEmpty :: Stack a ()
nonEmpty = empty >>= flip when (fail "Stack empty")</langsyntaxhighlight>
 
=={{header|JIcon}} and {{header|Unicon}}==
Stacks (and double ended queues) are built into Icon and Unicon as part of normal list access. In addition to 'push' and 'pop', there are the functions 'put', 'get' (alias for pop), 'pull', list element addressing, and list sectioning (like sub-strings).
Unicon extended 'insert' and 'delete' to work with lists.
The programmer is free to use any or all of the list processing functions on any problem.
The following illustrates typical stack usage:
<syntaxhighlight lang="icon">procedure main()
stack := [] # new empty stack
push(stack,1) # add item
push(stack,"hello",table(),set(),[],5) # add more items of mixed types in order left to right
y := top(stack) # peek
x := pop(stack) # remove item
write("The stack is ",if isempty(stack) then "empty" else "not empty")
end
 
procedure isempty(x) #: test if a datum is empty, return the datum or fail (task requirement)
<lang J>create_stack_=: monad def 'stack=:$0'
if *x = 0 then return x # in practice just write *x = 0 or *x ~= 0 for is/isn't empty
push_stack_=: monad def 'stack=:stack,y'
end
pop_stack_=: monad def 'r[ stack=:}:stack[ r=.{:stack'
empty_stack_=: monad def '0=#stack'</lang>
 
procedure top(x) #: return top element w/o changing stack
Example use:
return x[1] # in practice, just use x[1]
end</syntaxhighlight>
 
=={{header|Io}}==
<lang J> example=:conew~'stack'</lang>
aside from using built-in lists, a stack can be created using nodes like so:
<lang J> push__example 0</lang>
<syntaxhighlight lang="io">Node := Object clone do(
<lang J> pop__example 0</lang>
next := nil
<lang J> empty__example 0</lang>
obj := nil
)
 
Stack := Object clone do(
pop and empty ignore their arguments. In this implementation, push returns the new state of the stack.
node := nil
pop := method(
obj := node obj
node = node next
obj
)
push := method(obj,
nn := Node clone
nn obj = obj
nn next = self node
self node = nn
)
)</syntaxhighlight>
 
=={{header|Ioke}}==
<syntaxhighlight lang="ioke">Stack = Origin mimic do(
initialize = method(@elements = [])
pop = method(@elements pop!)
empty = method(@elements empty?)
push = method(element, @elements push!(element))
)</syntaxhighlight>
 
=={{header|IS-BASIC}}==
<syntaxhighlight lang="is-basic">100 LET N=255 ! Size of stack
110 NUMERIC STACK(1 TO N)
120 LET PTR=1
130 DEF PUSH(X)
140 IF PTR>N THEN
150 PRINT "Stack is full.":STOP
160 ELSE
170 LET STACK(PTR)=X:LET PTR=PTR+1
180 END IF
190 END DEF
200 DEF POP
210 IF PTR=1 THEN
220 PRINT "Stack is empty.":STOP
230 ELSE
240 LET PTR=PTR-1:LET POP=STACK(PTR)
250 END IF
260 END DEF
270 DEF EMPTY
280 LET PTR=1
290 END DEF
300 DEF TOP=STACK(PTR-1)
310 CALL PUSH(3):CALL PUSH(5)
320 PRINT POP+POP</syntaxhighlight>
 
=={{header|J}}==
<syntaxhighlight lang="j">stack=: ''
push=: monad def '0$stack=:stack,y'
pop=: monad def 'r[ stack=:}:stack[ r=.{:stack'
empty=: monad def '0=#stack'</syntaxhighlight>
Example use:
<syntaxhighlight lang="j"> push 9
 
pop ''
9
empty ''
1</syntaxhighlight>
pop and empty ignore their arguments. In this implementation. push returns an empty list.
 
=={{header|Java}}==
The collections framework includes a Stack class. Let's test it:
<lang java>
<syntaxhighlight lang="java">import java.util.Stack;
public class Stack
 
{
public class StackTest {
public static void main( final String[] args ) {
final Stack<String> stack = new Stack<String>();
 
System.out.println( "New stack empty? " + stack.empty() );
 
stack.push( "There can be only one" );
System.out.println( "Pushed stack empty? " + stack.empty() );
System.out.println( "Popped single entry: " + stack.pop() );
 
stack.push( "First" );
stack.push( "Second" );
System.out.println( "Popped entry should be second: " + stack.pop() );
 
// Popping an empty stack will throw...
stack.pop();
stack.pop();
}
}</syntaxhighlight>
{{out}}
<pre>New stack empty? true
Pushed stack empty? false
Popped single entry: There can be only one
Popped entry should be second: Second
Exception in thread "main" java.util.EmptyStackException
at java.util.Stack.peek(Stack.java:85)
at java.util.Stack.pop(Stack.java:67)
at StackTest.main(StackTest.java:21)</pre>
 
Alternatively, you might implement a stack yourself...
<syntaxhighlight lang="java">public class Stack{
private Node first = null;
public boolean isEmpty (){
return (first == null);
}
public Object Pop() {
if (first == nullisEmpty())
throw new Exception("Can't Pop from an empty Stack.");
else {
Object temp = first.Valuevalue;
first = first.Nextnext;
return temp;
}
}
public void Push(Object o) {
first = new Node(o, first);
}
class Node {
public Node Nextnext;
public Object Valuevalue;
public Node(Object value) {
this(value, null);
}
public Node(Object value, Node next) {
Nextthis.next = next;
Valuethis.value = value;
}
}
}</syntaxhighlight>
}
</lang>
 
{{works with|Java|1.5}}
<syntaxhighlight lang="java5">public class Stack<T>{
 
<lang java5>
public class Stack<T>
{
private Node first = null;
public boolean isEmpty (){
return (first == null);
}
public T Pop() {
if (first == nullisEmpty())
throw new Exception("Can't Pop from an empty Stack.");
else {
T temp = first.Valuevalue;
first = first.Nextnext;
return temp;
}
}
public void Push(T o) {
first = new Node(o, first);
}
class Node {
public Node Nextnext;
public T Valuevalue;
public Node(T value) {
this(value, null);
}
public Node(T value, Node next) {
Nextthis.next = next;
Valuethis.value = value;
}
}
}</syntaxhighlight>
}
</lang>
 
=={{header|JavaScript}}==
The built-in Array class already has stack primitives.
<langsyntaxhighlight lang="javascript">var stack = [];
var stack = [];
stack.push(1)
stack.push(2,3);
print(stack.pop()); // 3
print(stack.length); // 2, stack empty if 0</syntaxhighlight>
Here's a constructor that wraps the array:
</lang>
<syntaxhighlight lang="javascript">function Stack() {
this.data = new Array();
 
this.push = function(element) {this.data.push(element)}
this.pop = function() {return this.data.pop()}
this.empty = function() {return this.data.length == 0}
this.peek = function() {return this.data[this.data.length - 1]}
}</syntaxhighlight>
Here's an example using the revealing module pattern instead of prototypes.
<syntaxhighlight lang="javascript">
function makeStack() {
var stack = [];
 
var popStack = function () {
return stack.pop();
};
var pushStack = function () {
return stack.push.apply(stack, arguments);
};
var isEmpty = function () {
return stack.length === 0;
};
var peekStack = function () {
return stack[stack.length-1];
};
return {
pop: popStack,
push: pushStack,
isEmpty: isEmpty,
peek: peekStack,
top: peekStack
};
}
</syntaxhighlight>
 
=={{header|Jsish}}==
From Javascript entry. Being ECMAScript, Jsi supports stack primitives as part of the Array methods.
<syntaxhighlight lang="javascript">/* Stack, is Jsish */
var stack = [];
puts('depth:', stack.length);
 
stack.push(42);
stack.push('abc');
puts('depth:', stack.length);
 
puts('popped:', stack.pop());
if (stack.length) printf('not '); printf('empty\n');
puts('top:', stack[stack.length-1]);
puts('popped:', stack.pop());
if (stack.length) printf('not '); printf('empty\n');
 
puts('depth:', stack.length);</syntaxhighlight>
 
{{out}}
<pre>prompt$ jsish stack.jsi
depth: 0
depth: 2
popped: abc
not empty
top: 42
popped: 42
empty
depth: 0</pre>
 
=={{header|jq}}==
For most purposes, jq's arrays can be used for stacks if needed, without much further ado.
However, since the present task requires the definition of special stack-oriented operations,
we shall start with the following definitions:
<syntaxhighlight lang=jq>
# create a Stack
def Stack: {stack: []};
 
# check an object is a Stack
def isStack:
type == "object" and has("stack") and (.stack|type) == "array";
 
def pop:
if .stack|length == 0 then "pop: stack is empty" | error
else {stack: .stack[1:], item: .stack[0]]
end;
 
def push($x):
.stack = [$x] + .stack | .item = null;
 
def size:
.stack | length;
 
def isEmpty:
size == 0;
</syntaxhighlight>
 
Depending on context, additional code to check for or to enforce
type discipline could be added, but is omitted for simplicity here.
If using the C implementation of jq, the function names could also
be prefixed with "Stack::" to distinguish them as stack-oriented
operations.
 
For some purposes, this approach may be sufficient, but it can easily
become cumbersome if a sequence of operations must be performed
while also producing outputs that reflect intermediate states.
 
Suppose for example that we wish to create a stack, push some value,
and then pop the stack, obtaining the popped value as the final
result. This could be accomplished by the pipe:
<syntaxhighlight lang=jq>
Stack | push(3) | pop | .item
</syntaxhighlight>
 
Now suppose we also wish to record the size of the stack after each of these three operations.
One way to do this would be to write:
<syntaxhighlight lang=jq>
Stack
| size, (push(3)
| size, (pop
| size, .item ))
</syntaxhighlight>
 
Unfortunately this approach is error-prone and can quickly become tedious, so
we introduce an "observer" function that can "observe" intermediate states following any operation.
With observer/2 as defined below, we can instead write:
<syntaxhighlight lang=jq>
null
| observe(Stack; size)
| observe(push(3); size)
| observe(pop; size)
| .emit, item
</syntaxhighlight>
 
The idea is that each call to `observe` updates the "emit" slot, so that
all the accumulated messages are available at any point in the pipeline.
 
<syntaxhighlight lang=jq>
# Input: an object
# Output: the updated object with .emit filled in from `update|emit`.
# `emit` may produce a stream of values, which need not be strings.
def observe(update; emit):
def s(stream): reduce stream as $_ (null;
if $_ == null then .
elif . == null then "\($_)"
else . + "\n\($_)"
end);
.emit as $x
| update
| .emit = s($x // null, emit);
</syntaxhighlight>
 
 
=={{header|Julia}}==
{{works with|Julia|0.6}}
 
The built-in <code>Array</code> class already has efficient (linear amortized time) stack primitives.
<syntaxhighlight lang="julia">stack = Int[] # []
@show push!(stack, 1) # [1]
@show push!(stack, 2) # [1, 2]
@show push!(stack, 3) # [1, 2, 3]
@show pop!(stack) # 3
@show length(stack) # 2
@show empty!(stack) # []
@show isempty(stack) # true</syntaxhighlight>
 
=={{header|K}}==
<syntaxhighlight lang="k">stack:()
push:{stack::x,stack}
pop:{r:*stack;stack::1_ stack;r}
empty:{0=#stack}
 
/example:
stack:()
push 3
stack
,3
push 5
stack
5 3
pop[]
5
stack
,3
empty[]
0
pop[]
3
stack
!0
empty[]
1
</syntaxhighlight>
 
=={{header|Kotlin}}==
Rather than use the java.util.Stack<E> class, we will write our own simple Stack<E> class for this task:
<syntaxhighlight lang="scala">// version 1.1.2
 
class Stack<E> {
private val data = mutableListOf<E>()
 
val size get() = data.size
 
val empty get() = size == 0
 
fun push(element: E) = data.add(element)
 
fun pop(): E {
if (empty) throw RuntimeException("Can't pop elements from an empty stack")
return data.removeAt(data.lastIndex)
}
 
val top: E
get() {
if (empty) throw RuntimeException("Empty stack can't have a top element")
return data.last()
}
 
fun clear() = data.clear()
 
override fun toString() = data.toString()
}
 
fun main(args: Array<String>) {
val s = Stack<Int>()
(1..5).forEach { s.push(it) }
println(s)
println("Size of stack = ${s.size}")
print("Popping: ")
(1..3).forEach { print("${s.pop()} ") }
println("\nRemaining on stack: $s")
println("Top element is now ${s.top}")
s.clear()
println("After clearing, stack is ${if(s.empty) "empty" else "not empty"}")
try {
s.pop()
}
catch (e: Exception) {
println(e.message)
}
}</syntaxhighlight>
 
{{out}}
<pre>
[1, 2, 3, 4, 5]
Size of stack = 5
Popping: 5 4 3
Remaining on stack: [1, 2]
Top element is now 2
After clearing, stack is empty
Can't pop elements from an empty stack
</pre>
 
=={{header|Lambdatalk}}==
The APIs of stacks and queues are built on lambdatalk array primitives, [A.new, A.disp, A.join, A.split, A.array?, A.null?, A.empty?, A.in?, A.equal?, A.length, A.get, A.first, A.last, A.rest, A.slice, A.duplicate, A.reverse, A.concat, A.map, A.set!, A.addlast!, A.sublast!, A.addfirst!, A.subfirst!, A.reverse!, A.sort!, A.swap!, A.lib]. Note that the [A.addlast!, A.sublast!, A.addfirst!, A.subfirst!] primitives are the standard [push!, shift!, pop!, unshift!] ones.
 
<syntaxhighlight lang="scheme">
{def stack.add
{lambda {:v :s}
{let { {_ {A.addfirst! :v :s}}}
} ok}}
-> stack.add
 
{def stack.get
{lambda {:s}
{let { {:v {A.first :s}}
{_ {A.subfirst! :s}}
} :v}}}
-> stack.get
 
{def stack.peek
{lambda {:s}
{A.first :s}}}
-> stack.peek
 
{def stack.empty?
{lambda {:s}
{A.empty? :s}}}
-> stack.empty?
 
{def S {A.new}} -> S []
{stack.add 1 {S}} -> ok [1]
{stack.add 2 {S}} -> ok [2,1]
{stack.add 3 {S}} -> ok [3,2,1]
{stack.empty? {S}} -> false
{stack.get {S}} -> 3 [2,1]
{stack.add 4 {S}} -> ok [4,2,1]
{stack.peek {S}} -> 4 [4,2,1]
{stack.get {S}} -> 4 [2,1]
{stack.get {S}} -> 2 [1]
{stack.get {S}} -> 1 []
{stack.get {S}} -> undefined
{stack.empty? {S}} -> true
 
</syntaxhighlight>
 
=={{header|lang5}}==
<syntaxhighlight lang="lang5">: cr "\n" . ;
: empty? dup execute length if 0 else -1 then swap drop ;
: pop dup execute length 1 - extract swap drop ;
: push dup execute rot append over ;
: s. stack execute . ;
 
[] '_ set
: stack '_ ;
stack # local variable
1 swap push set
2 swap push set s. cr # [ 1 2 ]
pop . s. cr # 2 [ 1 ]
pop drop
empty? . # -1</syntaxhighlight>
 
=={{header|Lasso}}==
Lasso Arrays natively supports push and pop.
 
<syntaxhighlight lang="lasso">local(a) = array
 
#a->push('a')
#a->push('b')
#a->push('c')
 
#a->pop // c
#a->pop // b
#a->pop // a
#a->pop // null</syntaxhighlight>
 
=={{header|Liberty BASIC}}==
<syntaxhighlight lang="lb">
global stack$
stack$=""
 
randomize .51
for i = 1 to 10
if rnd(1)>0.5 then
print "pop => ";pop$()
else
j=j+1
s$ = chr$(j + 64)
print "push ";s$
call push s$
end if
next
 
print
print "Clean-up"
do
print "pop => ";pop$()
loop while not(empty())
print "Stack is empty"
 
end
 
'------------------------------------
sub push s$
stack$=s$+"|"+stack$ 'stack
end sub
 
function pop$()
if stack$="" then pop$="*EMPTY*": exit function
pop$=word$(stack$,1,"|")
stack$=mid$(stack$,instr(stack$,"|")+1)
end function
 
function empty()
empty =(stack$="")
end function
</syntaxhighlight>
 
=={{header|Lingo}}==
<syntaxhighlight lang="lingo">-- parent script "Stack"
 
property _tos
 
on push (me, data)
me._tos = [#data:data, #next:me._tos]
end
 
on pop (me)
if voidP(me._tos) then return VOID
data = me._tos.data
me._tos = me._tos.next
return data
end
 
on peek (me)
if voidP(me._tos) then return VOID
return me._tos.data
end
 
on empty (me)
return voidP(me.peek())
end</syntaxhighlight>
 
=={{header|Logo}}==
[[UCB Logo]] has built-in methods for treating lists as stacks. Since they are destructive, they take the name of the stack rather than the list itself.
<syntaxhighlight lang="logo">make "stack []
push "stack 1
push "stack 2
push "stack 3
print pop "stack ; 3
print empty? :stack ; false</syntaxhighlight>
 
=={{header|OCamlLogtalk}}==
A stack can be trivially represented using the built-in representation for lists:
<syntaxhighlight lang="logtalk">
:- object(stack).
 
:- public(push/3).
push(Element, Stack, [Element| Stack]).
 
:- public(pop/3).
pop([Top| Stack], Top, Stack).
 
:- public(empty/1)
empty([]).
 
:- end_object.
</syntaxhighlight>
 
=={{header|LOLCODE}}==
{{trans|UNIX Shell}}
<syntaxhighlight lang="lolcode">HAI 2.3
HOW IZ I Init YR Stak
Stak HAS A Length ITZ 0
IF U SAY SO
 
HOW IZ I Push YR Stak AN YR Value
Stak HAS A SRS Stak'Z Length ITZ Value
Stak'Z Length R SUM OF Stak'Z Length AN 1
IF U SAY SO
 
HOW IZ I Top YR Stak
FOUND YR Stak'Z SRS DIFF OF Stak'Z Length AN 1
IF U SAY SO
 
HOW IZ I Pop YR Stak
I HAS A Top ITZ I IZ Top YR Stak MKAY
Stak'Z Length R DIFF OF Stak'Z Length AN 1
FOUND YR Top
IF U SAY SO
 
HOW IZ I Empty YR Stak
FOUND YR BOTH SAEM 0 AN Stak'Z Length
IF U SAY SO
 
I HAS A Stak ITZ A BUKKIT
I IZ Init YR Stak MKAY
I IZ Push YR Stak AN YR "Fred" MKAY
I IZ Push YR Stak AN YR "Wilma" MKAY
I IZ Push YR Stak AN YR "Betty" MKAY
I IZ Push YR Stak AN YR "Barney" MKAY
 
IM IN YR Loop UPPIN YR Dummy TIL I IZ Empty YR Stak MKAY
VISIBLE I IZ Pop YR Stak MKAY
IM OUTTA YR Loop
 
KTHXBYE</syntaxhighlight>
 
{{Out}}
<pre>Barney
Betty
Wilma
Fred</pre>
 
=={{header|Lua}}==
Tables have stack primitives by default:
<syntaxhighlight lang="lua">stack = {}
table.insert(stack,3)
print(table.remove(stack)) --> 3</syntaxhighlight>
 
=={{header|M2000 Interpreter}}==
A Stack object can be used as LIFO or FIFO. Push statement push to top of stack. Read pop a value to a variable from top of stack. StackItem(1) read top item without modified stack. Data statement append items to bottom.
<syntaxhighlight lang="m2000 interpreter">
Module Checkit {
a=Stack
Stack a {
Push 100, 200, 300
}
Print StackItem(a, 1)=300
Stack a {
Print StackItem(1)=300
While not empty {
Read N
Print N
}
}
}
Checkit
</syntaxhighlight>
 
Every module and function has a "current" stack. Number is a read only variable, which pop a value from current stack (or raise error if not number is in top of stack).
 
User functions get a new stack, and drop it at return. Modules take parent stack, and return stack to parent. So a Module can return values too. In M2000 a call happen without checkig signatures (except for special events calls). We have to leave stack at a proper state, when return from a module.
Return/Execution stack is hidden and different from stack of values.
 
<syntaxhighlight lang="m2000 interpreter">
Module Checkit {
Read a, b
Print a, b
}
\\ add parameters in a FIFO, and this FIFO merged to current stack
Push 100
Checkit 10, 20
Print StackItem(1)=100
Module Checkit {
Read a, b
Print a=20, b=100
}
Checkit 20
 
Function alfa {
k=0
n=0
while not empty {
k+=number
n++
}
if n=0 then Error "No parameters found"
=k/n
}
 
Print alfa(1,2,3,4)=2.5
 
</syntaxhighlight>
 
=={{header|Maple}}==
<syntaxhighlight lang="maple">with(stack): # load the package, to allow use of short command names
 
s := stack:-new(a, b):
 
push(c, s):
 
# The following statements terminate with a semicolon and print output.
top(s);
pop(s);
pop(s);
empty(s);
pop(s);
empty(s);</syntaxhighlight>
{{out}}
<pre> c
 
c
 
b
 
false
 
a
 
true</pre>
 
=={{header|Mathematica}}/{{header|Wolfram Language}}==
<syntaxhighlight lang="mathematica">EmptyQ[a_] := If[Length[a] == 0, True, False]
SetAttributes[Push, HoldAll];[a_, elem_] := AppendTo[a, elem]
SetAttributes[Pop, HoldAllComplete];
Pop[a_] := If[EmptyQ[a], False, b = Last[a]; Set[a, Most[a]]; b]
Peek[a_] := If[EmptyQ[a], False, Last[a]]
 
Example use:
stack = {};Push[stack, 1]; Push[stack, 2]; Push[stack, 3]; Push[stack, 4];
Peek[stack]
->4
Pop[stack]
->4
Peek[stack]
->3</syntaxhighlight>
 
=={{header|MATLAB}} / {{header|Octave}}==
Here is a simple implementation of a stack, that works in Matlab and Octave. It is closely related to the queue/fifo example.
<syntaxhighlight lang="matlab">mystack = {};
% push
mystack{end+1} = x;
 
%pop
x = mystack{end}; mystack{end} = [];
 
%peek,top
x = mystack{end};
 
% empty
isempty(mystack)</syntaxhighlight>
Below is another solution, that encapsulates the fifo within the object-orientated "class" elements supported by Matlab. The given implementation is exactly the same as the MATLAB FIFO example, except that the "push()" function is modified to add stuff to the end of the queue instead of the beginning. This is a naive implementation, for rigorous applications this should be modified to initialize the LIFO to a buffered size, so that the "pop()" and "push()" functions don't resize the cell array that stores the LIFO's elements, every time they are called.
 
To use this implementation you must save this code in a MATLAB script file named "LIFOQueue.m" which must be saved in a folder named @LIFOQueue in your MATLAB directory.
<syntaxhighlight lang="matlab">%This class impliments a standard LIFO queue.
classdef LIFOQueue
properties
queue
end
methods
%Class constructor
function theQueue = LIFOQueue(varargin)
if isempty(varargin) %No input arguments
%Initialize the queue state as empty
theQueue.queue = {};
elseif (numel(varargin) > 1) %More than 1 input arg
%Make the queue the list of input args
theQueue.queue = varargin;
elseif iscell(varargin{:}) %If the only input is a cell array
%Make the contents of the cell array the elements in the queue
theQueue.queue = varargin{:};
else %There is one input argument that is not a cell
%Make that one arg the only element in the queue
theQueue.queue = varargin;
end
end
%push() - pushes a new element to the end of the queue
function push(theQueue,varargin)
if isempty(varargin)
theQueue.queue(end+1) = {[]};
elseif (numel(varargin) > 1) %More than 1 input arg
%Make the queue the list of input args
theQueue.queue( end+1:end+numel(varargin) ) = varargin;
elseif iscell(varargin{:}) %If the only input is a cell array
%Make the contents of the cell array the elements in the queue
theQueue.queue( end+1:end+numel(varargin{:}) ) = varargin{:};
else %There is one input argument that is not a cell
%Make that one arg the only element in the queue
theQueue.queue{end+1} = varargin{:};
end
%Makes changes to the queue permanent
assignin('caller',inputname(1),theQueue);
end
%pop() - pops the first element off the queue
function element = pop(theQueue)
if empty(theQueue)
error 'The queue is empty'
else
%Returns the first element in the queue
element = theQueue.queue{end};
%Removes the first element from the queue
theQueue.queue(end) = [];
%Makes changes to the queue permanent
assignin('caller',inputname(1),theQueue);
end
end
%empty() - Returns true if the queue is empty
function trueFalse = empty(theQueue)
trueFalse = isempty(theQueue.queue);
end
end %methods
end</syntaxhighlight>
Sample Usage:
<syntaxhighlight lang="matlab">>> myLIFO = LIFOQueue(1,'fish',2,'fish','red fish','blue fish')
myLIFO =
LIFOQueue
 
>> myLIFO.pop()
 
ans =
 
blue fish
 
>> myLIFO.push('Cat Fish')
>> myLIFO.pop()
 
ans =
 
Cat Fish
 
>> myLIFO.pop()
 
ans =
 
red fish
 
>> empty(myLIFO)
 
ans =
 
0</syntaxhighlight>
 
=={{header|Maxima}}==
<syntaxhighlight lang="maxima">/* lists can be used as stacks; Maxima provides pop and push */
 
load(basic)$
 
a: []$
push(25, a)$
push(7, a)$
pop(a);
 
emptyp(a);
length(a);</syntaxhighlight>
 
=={{header|Mercury}}==
 
Efficient, generic stacks are provided as part of the standard library in Mercury. For sake of illustration, here is how a simple stack could be implemented.
 
<syntaxhighlight lang="mercury">:- module sstack.
 
:- interface.
 
% We're going to call the type sstack (simple stack) because we don't want to get it
% accidentally confused with the official stack module in the standard library.
:- type sstack(T).
 
:- func sstack.new = sstack(T).
:- pred sstack.is_empty(sstack(T)::in) is semidet.
:- func sstack.push(sstack(T), T) = sstack(T).
:- pred sstack.pop(T::out, sstack(T)::in, sstack(T)::out) is semidet.
 
:- implementation.
 
:- import_module list.
 
:- type sstack(T)
---> sstack(list(T)).
 
sstack.new = sstack([]).
 
sstack.is_empty(sstack([])).
 
sstack.push(Stack0, Elem) = Stack1 :-
Stack0 = sstack(Elems),
Stack1 = sstack([Elem | Elems]).
 
sstack.pop(Elem, !Stack) :-
!.Stack = sstack([Elem | Elems]),
!:Stack = sstack(Elems).
 
:- end_module sstack.</syntaxhighlight>
 
It should be noted that this is purely an illustrative example of a very simple stack.
A real implementation would have predicate (:- pred) versions of the functions (:- func), for example, for consistency's sake with either the functions implemented in terms of the predicates or vice versa. [http://www.mercurylang.org/information/doc-release/mercury_library/stack.html#stack The real library implementation] also features more functionality including both semi-deterministic and deterministic versions of some functions/predicates as well as the ability to push a list of values in one operation.
 
Some of the implementation decisions above need an explanation.
new/0 and push/2 were implemented as functions both for pedagogical reasons (a desire to show function syntax) and because they are a natural fit for functional thought: 0 or more inputs, one output, deterministic.
is_empty/1 was implemented as a predicate because it's a single, simple succeed/fail test which is precisely what a predicate is in logic.
pop/3 was implemented as a predicate because it has two outputs (the element and the new stack) ''and'' because it is semi-deterministic (it will fail if the stack is empty).
 
Note also that while pop/3 has three parameters, the function implementation looks like it has two. This is because the !Stack "parameter" is actually a ''pair'' of parameters using Mercury's state variable notation. !Stack is, in effect, two variables: !.Stack and !:Stack, input and output respectively. Using state variable notation here is a bit of overkill but again was brought in for pedagogical reasons to show the syntax.
 
=={{header|MIPS Assembly}}==
 
<syntaxhighlight lang="mips">addi sp,sp,-4
sw t0,0(sp) ;push
 
lw t0,0(sp)
addi sp,sp,4 ;pop
 
lw t0,0(sp) ;top</syntaxhighlight>
 
"Empty" requires you to know the starting value of <code>SP</code>. Since it's hardware-dependent, there's no one answer for this part of the task.
 
=={{header|MiniScript}}==
<syntaxhighlight lang="miniscript">// Note in Miniscript, a value of zero is false,
// and any other number is true.
// therefore the .len function works as the inverse of a .empty function
stack = [2, 4, 6]
stack.push 8
print "Stack is " + stack
print "Adding '9' to stack " + stack.push(9)
print "Top of stack is " + stack.pop
print "Stack is " + stack
if stack.len then
print "Stack is not empty"
else
print "Stack is empty"
end if</syntaxhighlight>
{{out}}
<pre>
Stack is [2, 4, 6, 8]
Adding '9' to stack [2, 4, 6, 8, 9]
Top of stack is 9
Stack is [2, 4, 6, 8]
Stack is not empty
</pre>
 
=={{header|Nanoquery}}==
<syntaxhighlight lang="nanoquery">class Stack
declare internalList
 
// constructor
def Stack()
internalList = list()
end
 
def push(val)
internalList.append(val)
end
 
def pop()
val = internalList[int(len($internalList) - 1)]
internalList.remove(val)
 
return val
end
 
def empty()
return len(internalList) = 0
end
end</syntaxhighlight>
 
=={{header|Nemerle}}==
Mutable stacks are available in <tt>System.Collections</tt>, <tt>System.Collections.Generic</tt> and <tt>Nemerle.Collections</tt> depending on what functionality beyond the basics you want. An immutable stack could be implemented fairly easily, as, for example, this quick and dirty list based implementation.
<syntaxhighlight lang="nemerle">public class Stack[T]
{
private stack : list[T];
public this()
{
stack = [];
}
public this(init : list[T])
{
stack = init;
}
public Push(item : T) : Stack[T]
{
Stack(item::stack)
}
public Pop() : T * Stack[T]
{
(stack.Head, Stack(stack.Tail))
}
public Peek() : T
{
stack.Head
}
public IsEmpty() : bool
{
stack.Length == 0
}
}</syntaxhighlight>
 
=={{header|NetRexx}}==
<syntaxhighlight lang="netrexx">/* NetRexx ************************************************************
* 13.08.2013 Walter Pachl translated from REXX version 2
**********************************************************************/
options replace format comments java crossref savelog symbols nobinary
 
stk = create_stk
 
say push(stk,123) 'from push'
say empty(stk)
say peek(stk) 'from peek'
say pull(stk) 'from pull'
say empty(stk)
Say pull(stk) 'from pull'
 
method create_stk static returns Rexx
stk = ''
stk[0] = 0
return stk
 
method push(stk,v) static
stk[0]=stk[0]+1
stk[stk[0]]=v
Return v
 
method peek(stk) static
x=stk[0]
If x=0 Then
Return 'stk is empty'
Else
Return stk[x]
 
method pull(stk) static
x=stk[0]
If x=0 Then
Return 'stk is empty'
Else Do
stk[0]=stk[0]-1
Return stk[x]
End
 
method empty(stk) static
Return stk[0]=0</syntaxhighlight>
{{out}}
<pre>
123 from push
0
123 from peek
123 from pull
1
stk is empty from pull
</pre>
 
=={{header|Nim}}==
In Nim, the sequences offer all the functionalities of a stack. Procedure <code>add</code> appends an item at the end, procedure <code>pop</code> returns the last element and removes it from the sequence. And it’s easy to check if if the sequence is empty with the procedure <code>len</code> which returns its length.
 
If we want a stack type limited to the four or five functions of the task, it is possible to define a distinct generic type <code>Stack[T]</code> derived from <code>seq[T]</code>. The code will be typically as follows. Note that we have defined a procedure <code>top</code> to get the value of the top item, another <code>mtop</code> to get a mutable reference to the top item and also a procedure <code>mtop=</code> to assign directly a value to the top item.
 
<syntaxhighlight lang="nim">type Stack[T] = distinct seq[T]
 
func initStack[T](initialSize = 32): Stack[T] =
Stack[T](newSeq[T](initialSize))
 
func isEmpty[T](stack: Stack[T]): bool =
seq[T](stack).len == 0
 
func push[T](stack: var Stack[T]; item: sink T) =
seq[T](stack).add(item)
 
func pop[T](stack: var Stack[T]): T =
if stack.isEmpty:
raise newException(IndexDefect, "stack is empty.")
seq[T](stack).pop()
 
func top[T](stack: Stack[T]): T =
if stack.isEmpty:
raise newException(IndexDefect, "stack is empty.")
seq[T](stack)[^1]
 
func mtop[T](stack: var Stack[T]): var T =
if stack.isEmpty:
raise newException(IndexDefect, "stack is empty.")
seq[T](stack)[^1]
 
func `mtop=`[T](stack: var Stack[T]; value: T) =
if stack.isEmpty:
raise newException(IndexDefect, "stack is empty.")
seq[T](stack)[^1] = value
 
when isMainModule:
 
var s = initStack[int]()
s.push 2
echo s.pop
s.push 3
echo s.top
s.mtop += 1
echo s.top
s.mtop = 5
echo s.top</syntaxhighlight>
 
{{out}}
<pre>2
3
4
5</pre>
 
=={{header|Oberon-2}}==
{{Works with|oo2c version 2}}
<syntaxhighlight lang="oberon2">
MODULE Stacks;
IMPORT
Object,
Object:Boxed,
Out := NPCT:Console;
 
TYPE
Pool(E: Object.Object) = POINTER TO ARRAY OF E;
Stack*(E: Object.Object) = POINTER TO StackDesc(E);
StackDesc*(E: Object.Object) = RECORD
pool: Pool(E);
cap-,top: LONGINT;
END;
 
PROCEDURE (s: Stack(E)) INIT*(cap: LONGINT);
BEGIN
NEW(s.pool,cap);s.cap := cap;s.top := -1
END INIT;
 
PROCEDURE (s: Stack(E)) Top*(): E;
BEGIN
RETURN s.pool[s.top]
END Top;
 
PROCEDURE (s: Stack(E)) Push*(e: E);
BEGIN
INC(s.top);
ASSERT(s.top < s.cap);
s.pool[s.top] := e;
END Push;
 
PROCEDURE (s: Stack(E)) Pop*(): E;
VAR
resp: E;
BEGIN
ASSERT(s.top >= 0);
resp := s.pool[s.top];DEC(s.top);
RETURN resp
END Pop;
 
PROCEDURE (s: Stack(E)) IsEmpty(): BOOLEAN;
BEGIN
RETURN s.top < 0
END IsEmpty;
 
PROCEDURE (s: Stack(E)) Size*(): LONGINT;
BEGIN
RETURN s.top + 1
END Size;
 
PROCEDURE Test;
VAR
s: Stack(Boxed.LongInt);
BEGIN
s := NEW(Stack(Boxed.LongInt),100);
s.Push(NEW(Boxed.LongInt,10));
s.Push(NEW(Boxed.LongInt,100));
Out.String("size: ");Out.Int(s.Size(),0);Out.Ln;
Out.String("pop: ");Out.Object(s.Pop());Out.Ln;
Out.String("top: ");Out.Object(s.Top());Out.Ln;
Out.String("size: ");Out.Int(s.Size(),0);Out.Ln
END Test;
BEGIN
Test
END Stacks.
</syntaxhighlight>
{{out}}
<pre>
size: 2
pop: 100
top: 10
size: 1
</pre>
{{works with|AOS}}
<syntaxhighlight lang="oberon2">
MODULE Stacks; (** AUTHOR ""; PURPOSE ""; *)
 
IMPORT
Out := KernelLog;
 
TYPE
Object = OBJECT
END Object;
Stack* = OBJECT
VAR
top-,capacity-: LONGINT;
pool: POINTER TO ARRAY OF Object;
PROCEDURE & InitStack*(capacity: LONGINT);
BEGIN
SELF.capacity := capacity;
SELF.top := -1;
NEW(SELF.pool,capacity)
END InitStack;
PROCEDURE Push*(a:Object);
BEGIN
INC(SELF.top);
ASSERT(SELF.top < SELF.capacity,100);
SELF.pool[SELF.top] := a
END Push;
PROCEDURE Pop*(): Object;
VAR
r: Object;
BEGIN
ASSERT(SELF.top >= 0);
r := SELF.pool[SELF.top];
DEC(SELF.top);RETURN r
END Pop;
PROCEDURE Top*(): Object;
BEGIN
ASSERT(SELF.top >= 0);
RETURN SELF.pool[SELF.top]
END Top;
PROCEDURE IsEmpty*(): BOOLEAN;
BEGIN
RETURN SELF.top < 0
END IsEmpty;
END Stack;
BoxedInt = OBJECT
(Object)
VAR
val-: LONGINT;
 
PROCEDURE & InitBoxedInt*(CONST val: LONGINT);
BEGIN
SELF.val := val
END InitBoxedInt;
 
END BoxedInt;
 
PROCEDURE Test*;
VAR
s: Stack;
bi: BoxedInt;
obj: Object;
BEGIN
NEW(s,10); (* A new stack of ten objects *)
NEW(bi,100);s.Push(bi);
NEW(bi,102);s.Push(bi);
NEW(bi,104);s.Push(bi);
Out.Ln;
Out.String("Capacity:> ");Out.Int(s.capacity,0);Out.Ln;
Out.String("Size:> ");Out.Int(s.top + 1,0);Out.Ln;
obj := s.Pop(); obj := s.Pop();
WITH obj: BoxedInt DO
Out.String("obj:> ");Out.Int(obj.val,0);Out.Ln
ELSE
Out.String("Unknown object...");Out.Ln;
END (* with *)
END Test;
END Stacks.
</syntaxhighlight>
{{out}}
<pre>
Capacity:> 10
Size:> 3
obj:> 102
</pre>
 
=={{header|Objeck}}==
Class library support for Stack/IntStack/FloatStack
<syntaxhighlight lang="objeck">stack := IntStack->New();
stack->Push(13);
stack->Push(7);
(stack->Pop() + stack->Pop())->PrintLine();
stack->IsEmpty()->PrintLine();</syntaxhighlight>
 
=={{header|Objective-C}}==
Using a NSMutableArray:
<syntaxhighlight lang="objc">NSMutableArray *stack = [NSMutableArray array]; // creating
 
[stack addObject:value]; // pushing
 
id value = [stack lastObject];
[stack removeLastObject]; // popping
 
[stack count] == 0 // is empty?</syntaxhighlight>
 
=={{header|OCaml}}==
Implemented as a singly-linked list, wrapped in an object:
<syntaxhighlight lang="ocaml">exception Stack_empty
<lang ocaml>
exception Stack_empty
 
class ['a] stack =
Line 733 ⟶ 4,994:
 
method push x =
lst <- x::lst
 
method pop =
Line 743 ⟶ 5,004:
method is_empty =
lst = []
end</syntaxhighlight>
 
</lang>
=={{header|Oforth}}==
 
Stack is already defined at startup.
 
<syntaxhighlight lang="oforth">ListBuffer Class new: Stack
Stack method: push self add ;
Stack method: pop self removeLast ;
Stack method: top self last ;</syntaxhighlight>
 
Usage :
<syntaxhighlight lang="oforth">: testStack
| s |
Stack new ->s
s push(10)
s push(11)
s push(12)
s top println
s pop println
s pop println
s pop println
s isEmpty ifTrue: [ "Stack is empty" println ] ;</syntaxhighlight>
 
{{out}}
<pre>
12
12
11
10
Stack is empty
</pre>
 
=={{header|Ol}}==
Simplest stack can be implemented using 'cons' and 'uncons' primitives.
<syntaxhighlight lang="scheme">
(define stack #null)
(print "stack is: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* pushing 1")
(define stack (cons 1 stack))
(print "stack is: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* pushing 2")
(define stack (cons 2 stack))
(print "stack is: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* pushing 3")
(define stack (cons 3 stack))
(print "stack is: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* poping")
(define-values (value stack) (uncons stack #f))
(print "value: " value)
(print "stack: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* poping")
(define-values (value stack) (uncons stack #f))
(print "value: " value)
(print "stack: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* poping")
(define-values (value stack) (uncons stack #f))
(print "value: " value)
(print "stack: " stack)
(print "is stack empty: " (eq? stack #null))
 
(print "* poping")
(define-values (value stack) (uncons stack #f))
(print "value: " value)
(print "stack: " stack)
(print "is stack empty: " (eq? stack #null))
</syntaxhighlight>
{{out}}
<pre>
stack is: ()
is stack empty: #true
* pushing 1
stack is: (1)
is stack empty: #false
* pushing 2
stack is: (2 1)
is stack empty: #false
* pushing 3
stack is: (3 2 1)
is stack empty: #false
* poping
value: 3
stack: (2 1)
is stack empty: #false
* poping
value: 2
stack: (1)
is stack empty: #false
* poping
value: 1
stack: ()
is stack empty: #true
* poping
value: #false
stack: ()
is stack empty: #true
</pre>
 
 
But in real programs may be useful a more complex stack implementation based on coroutines (ol is a purely functional lisp, so it does not support mutators like 'set!').
 
<syntaxhighlight lang="scheme">
(fork-server 'stack (lambda ()
(let this ((me '()))
(let*((envelope (wait-mail))
(sender msg envelope))
(case msg
(['empty]
(mail sender (null? me))
(this me))
(['push value]
(this (cons value me)))
(['pop]
(cond
((null? me)
(mail sender #false)
(this me))
(else
(mail sender (car me))
(this (cdr me))))))))))
(define (push value)
(mail 'stack ['push value]))
(define (pop)
(await (mail 'stack ['pop])))
(define (empty)
(await (mail 'stack ['empty])))
 
(for-each (lambda (n)
(print "pushing " n)
(push n))
(iota 5 1)) ; '(1 2 3 4 5)
 
(let loop ()
(print "is stack empty: " (empty))
(unless (empty)
(begin
(print "popping value, got " (pop))
(loop))))
(print "done.")
</syntaxhighlight>
 
{{out}}
<pre>
pushing 1
pushing 2
pushing 3
pushing 4
pushing 5
is stack empty: #false
popping value, got 5
is stack empty: #false
popping value, got 4
is stack empty: #false
popping value, got 3
is stack empty: #false
popping value, got 2
is stack empty: #false
popping value, got 1
is stack empty: #true
done.
</pre>
 
=={{header|ooRexx}}==
The ooRexx queue class functions as a stack as well (it is a dequeue really).
<syntaxhighlight lang="oorexx">
stack = .queue~of(123, 234) -- creates a stack with a couple of items
stack~push("Abc") -- pushing
value = stack~pull -- popping
value = stack~peek -- peeking
-- the is empty test
if stack~isEmpty then say "The stack is empty"
</syntaxhighlight>
 
=={{header|OxygenBasic}}==
The real stack is freely available!
<syntaxhighlight lang="oxygenbasic">
function f()
sys a=1,b=2,c=3,d=4
push a
push b
push c
push d
print a "," b "," c "," d 'result 1,2,3,4
a=10
b=20
c=30
d=40
print a "," b "," c "," d 'result 10,20,30,40
pop a
pop b
pop c
pop d
print a "," b "," c "," d 'result 4,3,2,1
end function
 
f
</syntaxhighlight>
 
=={{header|Oz}}==
A thread-safe, list-based stack. Implemented as a module:
<syntaxhighlight lang="oz">functor
export
New
Push
Pop
Empty
define
fun {New}
{NewCell nil}
end
 
proc {Push Stack Element}
NewStack
%% Use atomic swap for thread safety
OldStack = Stack := NewStack
in
NewStack = Element|OldStack
end
 
proc {Pop Stack ?Result}
NewStack
%% Use atomic swap for thread safety
OldStack = Stack := NewStack
in
Result|NewStack = OldStack
end
fun {Empty Stack}
@Stack == nil
end
end</syntaxhighlight>
There is also a stack implementation in the [http://www.mozart-oz.org/home/doc/mozart-stdlib/adt/stack.html standard library].
 
=={{header|PARI/GP}}==
<syntaxhighlight lang="parigp">push(x)=v=concat(v,[x]);;
pop()={
if(#v,
my(x=v[#v]);
v=vecextract(v,1<<(#v-1)-1);
x
,
error("Stack underflow")
)
};
empty()=v==[];
peek()={
if(#v,
v[#v]
,
error("Stack underflow")
)
};</syntaxhighlight>
 
=={{header|Pascal}}==
This implements stacks of integers in standard Pascal (should work on all existing Pascal dialects).
<syntaxhighlight lang="pascal">{ tStack is the actual stack type, tStackNode a helper type }
<lang pascal>
{ tStack is the actual stack type, tStackNode a helper type }
type
pStackNode = ^tStackNode;
Line 773 ⟶ 5,295:
while stack.top <> nil do
begin
node := stack.top;
stack.top := stack.top^.next;
dispose(node);
end
end;
 
function StackIsEmpty(stack: tStack):Boolean;
begin
StackIsEmpty := stack.top = nil
Line 803 ⟶ 5,325:
PopFromStack := node^.data;
dispose(node);
end;</syntaxhighlight>
</lang>
 
=={{header|Perl}}==
Perl comes prepared to treat its listsarrays as stacks, giving us the push and pop functions for free. To add empty, we basically give a new name to "not":
<syntaxhighlight lang="perl">sub empty{ not @_ }</syntaxhighlight>
 
=={{header|Phix}}==
<lang perl>sub empty{ not @_ }</lang>
<!--<syntaxhighlight lang="phix">(phixonline)-->
<span style="color: #008080;">with</span> <span style="color: #008080;">javascript_semantics</span>
<span style="color: #000080;font-style:italic;">-- comparing a simple implementation against using the builtins:</span>
<span style="color: #004080;">sequence</span> <span style="color: #000000;">stack</span> <span style="color: #0000FF;">=</span> <span style="color: #0000FF;">{}</span>
<span style="color: #008080;">procedure</span> <span style="color: #000000;">push_</span><span style="color: #0000FF;">(</span><span style="color: #004080;">object</span> <span style="color: #000000;">what</span><span style="color: #0000FF;">)</span>
<span style="color: #000000;">stack</span> <span style="color: #0000FF;">=</span> <span style="color: #7060A8;">append</span><span style="color: #0000FF;">(</span><span style="color: #000000;">stack</span><span style="color: #0000FF;">,</span><span style="color: #000000;">what</span><span style="color: #0000FF;">)</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">procedure</span>
<span style="color: #008080;">function</span> <span style="color: #000000;">pop_</span><span style="color: #0000FF;">()</span>
<span style="color: #004080;">object</span> <span style="color: #000000;">what</span> <span style="color: #0000FF;">=</span> <span style="color: #000000;">stack</span><span style="color: #0000FF;">[$]</span>
<span style="color: #000000;">stack</span> <span style="color: #0000FF;">=</span> <span style="color: #000000;">stack</span><span style="color: #0000FF;">[</span><span style="color: #000000;">1</span><span style="color: #0000FF;">..$-</span><span style="color: #000000;">1</span><span style="color: #0000FF;">]</span>
<span style="color: #008080;">return</span> <span style="color: #000000;">what</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">function</span>
<span style="color: #008080;">function</span> <span style="color: #000000;">empty_</span><span style="color: #0000FF;">()</span>
<span style="color: #008080;">return</span> <span style="color: #7060A8;">length</span><span style="color: #0000FF;">(</span><span style="color: #000000;">stack</span><span style="color: #0000FF;">)=</span><span style="color: #000000;">0</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">function</span>
<span style="color: #0000FF;">?</span><span style="color: #000000;">empty_</span><span style="color: #0000FF;">()</span> <span style="color: #000080;font-style:italic;">-- 1</span>
<span style="color: #000000;">push_</span><span style="color: #0000FF;">(</span><span style="color: #000000;">5</span><span style="color: #0000FF;">)</span>
<span style="color: #0000FF;">?</span><span style="color: #000000;">empty_</span><span style="color: #0000FF;">()</span> <span style="color: #000080;font-style:italic;">-- 0</span>
<span style="color: #000000;">push_</span><span style="color: #0000FF;">(</span><span style="color: #000000;">6</span><span style="color: #0000FF;">)</span>
<span style="color: #0000FF;">?</span><span style="color: #000000;">pop_</span><span style="color: #0000FF;">()</span> <span style="color: #000080;font-style:italic;">-- 6</span>
<span style="color: #0000FF;">?</span><span style="color: #000000;">pop_</span><span style="color: #0000FF;">()</span> <span style="color: #000080;font-style:italic;">-- 5</span>
<span style="color: #0000FF;">?</span><span style="color: #000000;">empty_</span><span style="color: #0000FF;">()</span> <span style="color: #000080;font-style:italic;">-- 1</span>
<span style="color: #0000FF;">?</span><span style="color: #008000;">"===builtins==="</span>
<span style="color: #7060A8;">requires</span><span style="color: #0000FF;">(</span><span style="color: #008000;">"1.0.2"</span><span style="color: #0000FF;">)</span> <span style="color: #000080;font-style:italic;">-- (latest bugfixes, plus top renamed as peep, for p2js)</span>
<span style="color: #004080;">integer</span> <span style="color: #000000;">sid</span> <span style="color: #0000FF;">=</span> <span style="color: #7060A8;">new_stack</span><span style="color: #0000FF;">()</span>
<span style="color: #0000FF;">?</span><span style="color: #7060A8;">stack_empty</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">)</span> <span style="color: #000080;font-style:italic;">-- 1</span>
<span style="color: #7060A8;">push</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">,</span><span style="color: #000000;">5</span><span style="color: #0000FF;">)</span>
<span style="color: #0000FF;">?</span><span style="color: #7060A8;">stack_empty</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">)</span> <span style="color: #000080;font-style:italic;">-- 0</span>
<span style="color: #7060A8;">push</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">,</span><span style="color: #000000;">6</span><span style="color: #0000FF;">)</span>
<span style="color: #000080;font-style:italic;">--?peep(sid) -- 6 (leaving it there)</span>
<span style="color: #0000FF;">?</span><span style="color: #7060A8;">pop</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">)</span> <span style="color: #000080;font-style:italic;">-- 6</span>
<span style="color: #0000FF;">?</span><span style="color: #7060A8;">pop</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">)</span> <span style="color: #000080;font-style:italic;">-- 5</span>
<span style="color: #0000FF;">?</span><span style="color: #7060A8;">stack_empty</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sid</span><span style="color: #0000FF;">)</span> <span style="color: #000080;font-style:italic;">-- 1</span>
<!--</syntaxhighlight>-->
Note you get true/false rather than 1/0 under pwa/p2js (use printf(%t) for consistent results)
 
=={{header|PHP}}==
PHP arrays behave like a stack:
<syntaxhighlight lang="php">$stack = array();
 
empty( $stack ); // true
 
array_push( $stack, 1 ); // or $stack[] = 1;
array_push( $stack, 2 ); // or $stack[] = 2;
 
empty( $stack ); // false
 
echo array_pop( $stack ); // outputs "2"
echo array_pop( $stack ); // outputs "1"</syntaxhighlight>
 
=={{header|PicoLisp}}==
The built-in functions [http://software-lab.de/doc/refP.html#push push] and
[http://software-lab.de/doc/refP.html#pop pop] are used to maintain a stack (of any type).
<syntaxhighlight lang="picolisp">(push 'Stack 3)
(push 'Stack 2)
(push 'Stack 1)</syntaxhighlight>
<pre>: Stack
-> (1 2 3)
 
: (pop 'Stack)
-> 1
 
: Stack
-> (2 3)
 
: (set 'Stack) # empty
-> NIL
 
: Stack
-> NIL</pre>
 
=={{header|Pike}}==
Pike has a built in module ''ADT'' (Abstract Data Types) which among
other things contains a stack.
 
<syntaxhighlight lang="pike">
object s = ADT.Stack();
s->push("a");
s->push("b");
write("top: %O, pop1: %O, pop2: %O\n",
s->top(), s->pop(), s->pop());
s->reset(); // Empty the stack
</syntaxhighlight>
{{Out}}
<pre>
top: "b", pop1: "b", pop2: "a"
</pre>
 
=={{header|PL/I}}==
<syntaxhighlight lang="pl/i">/* Any controlled variable may behave as a stack. */
 
declare s float controlled;
 
/* to push a value on the stack. */
allocate s;
s = 10;
 
/* To pop a value from the stack. */
put (s);
free s;
 
/* to peek at the top of stack> */
put (s);
 
/* To see whether the stack is empty */
if allocation(s) = 0 then ...
 
/* Note: popping a value from the stack, or peeking, */
/* would usually require a check that the stack is not empty. */
 
/* Note: The above is a simple stack for S. */
/* S can be any kind of data structure, an array, etc. */
 
/* Example to push ten values onto the stack, and then to */
/* remove them. */
 
/* Push ten values, obtained from the input, onto the stack: */
declare S float controlled;
do i = 1 to 10;
allocate s;
get list (s);
end;
/* To pop those values from the stack: */
do while (allocation(s) > 0);
put skip list (s);
free s;
end;
/* The values are printed in the reverse order, of course. */</syntaxhighlight>
 
=={{header|PostScript}}==
{{libheader|initlib}}
<syntaxhighlight lang="postscript">% empty? is already defined.
/push {exch cons}.
/pop {uncons exch pop}.
[2 3 4 5 6] 1 push
= [1 2 3 4 5 6]
[1 2 3 4 5 6] pop
=[2 3 4 5 6]
[2 3 4 5 6] empty?
=false
[] empty?
=true</syntaxhighlight>
 
=={{header|PowerShell}}==
A new stack:
<syntaxhighlight lang="powershell">
$stack = New-Object -TypeName System.Collections.Stack
# or
$stack = [System.Collections.Stack] @()
</syntaxhighlight>
Push some stuff on the stack:
<syntaxhighlight lang="powershell">
1, 2, 3, 4 | ForEach-Object {$stack.Push($_)}
</syntaxhighlight>
Show stack as a string:
<syntaxhighlight lang="powershell">
$stack -join ", "
</syntaxhighlight>
{{Out}}
<pre>
4, 3, 2, 1
</pre>
Pop the top level of the stack:
<syntaxhighlight lang="powershell">
$stack.Pop()
</syntaxhighlight>
{{Out}}
<pre>
4
</pre>
Show stack as a string:
<syntaxhighlight lang="powershell">
$stack -join ", "
</syntaxhighlight>
{{Out}}
<pre>
3, 2, 1
</pre>
Get a copy of the top level of the stack:
<syntaxhighlight lang="powershell">
$stack.Peek()
</syntaxhighlight>
{{Out}}
<pre>
3
</pre>
The stack:
<syntaxhighlight lang="powershell">
$stack
</syntaxhighlight>
{{Out}}
<pre>
3
2
1
</pre>
 
=={{header|Prolog}}==
Prolog is a particularly silly language to implement stack functions in, as the built-in lists can be treated as stacks in an ad hoc manner. Nonetheless, in the name of completeness:
<syntaxhighlight lang="prolog">% push( ELEMENT, STACK, NEW )
 
<lang prolog>% push( ELEMENT, STACK, NEW )
% True if NEW is [ELEMENT|STACK]
push(ELEMENT,STACK,[ELEMENT|STACK]).
Line 824 ⟶ 5,547:
% empty( STACK )
% True if STACK is empty
empty([]).</langsyntaxhighlight>
 
=={{header|PythonPureBasic}}==
For LIFO function PureBasic normally uses linked lists.
Usage as described above could look like;
<syntaxhighlight lang="purebasic">Global NewList MyStack()
 
Procedure Push_LIFO(n)
{{works with|Python|2.5}}
FirstElement(MyStack())
InsertElement(MyStack())
MyStack() = n
EndProcedure
 
Procedure Pop_LIFO()
The faster and Pythonic way is using a deque (available from 2.4). A regular list is little slower.
If FirstElement(MyStack())
Topmost = MyStack()
DeleteElement(MyStack())
EndIf
ProcedureReturn Topmost
EndProcedure
 
Procedure Empty_LIFO()
<lang python>
Protected Result
from collections import deque
If ListSize(MyStack())=0
Result = #True
EndIf
ProcedureReturn Result
EndProcedure
 
Procedure Peek_LIFO()
If FirstElement(MyStack())
Topmost = MyStack()
EndIf
ProcedureReturn Topmost
EndProcedure
 
;---- Example of implementation ----
Push_LIFO(3)
Push_LIFO(1)
Push_LIFO(4)
While Not Empty_LIFO()
Debug Pop_LIFO()
Wend</syntaxhighlight>
 
{{out}}
<pre>
4
1
3
</pre>
 
=={{header|Python}}==
{{works with|Python|2.5}}
The faster and Pythonic way is using a deque (available from 2.4).
A regular list is a little slower.
<syntaxhighlight lang="python">from collections import deque
stack = deque()
stack.append(value) # pushing
value = stack.pop()
not stack # is empty?</syntaxhighlight>
</lang>
 
If you need to expose your stack to the world, you may want to create a simpler wrapper:
<syntaxhighlight lang="python">from collections import deque
 
<lang python>
from collections import deque
 
class Stack:
Line 853 ⟶ 5,618:
return self._items.pop()
def __nonzero__(self):
return bool(self._items)</syntaxhighlight>
</lang>
 
Here is a stack implemented as linked list - with the same list interface.
<syntaxhighlight lang="python">class Stack:
 
<lang python>
class Stack:
def __init__(self):
self._first = None
Line 870 ⟶ 5,631:
raise IndexError, "pop from empty stack"
value, self._first = self._first
return value</syntaxhighlight>
</lang>
 
Notes:
 
Using list interface - append, __nonzero__ make it easier to use, cleanup the client code, and allow changing the implementation later without affecting the client code. For example, instead of:
For example, instead of:
<lang python>
<syntaxhighlight lang="python">while not stack.empty():</syntaxhighlight>
</lang>
You can write:
<syntaxhighlight lang="python">while stack:</syntaxhighlight>
<lang python>
while stack:
</lang>
 
Quick testing show that deque is about 5 times faster then the wrapper linked list implementations. This may be important if your stack is used in tight loops.
 
=={{header|Quackery}}==
 
Quackery is a stack based language. In addition to ''the stack'' (i.e. the Quackery data stack) and the call stack, named ancillary stacks can be created with <code>[ stack ] is <name-of-stack></code>. Pushing to and popping from ancillary stacks is done with the words <code>put</code> and <code>take</code>. A word to test if an ancillary stack is empty can be defined as <code>[ size 1 = ] is isempty</code>. (The word <code>empty</code> already has a meaning in Quackery.) The word <code>share</code> returns the topmost element of an ancillary stack without changing the ancillary stack. Other ancillary stack operations are also available.
 
<syntaxhighlight lang="quackery">[ size 1 = ] is isempty ( s --> b )
 
[ stack ] is mystack ( --> s )
 
mystack isempty if [ say "mystack is empty" cr cr ]
23 mystack put
mystack share echo say " is on the top of mystack" cr cr
mystack mystack put ( you can put anything on an ancillary stack, even itself! )
mystack share echo say " is on the top of mystack" cr cr
mystack take echo say " has been removed from mystack" cr cr
mystack take echo say " has been removed from mystack" cr cr
mystack isempty if [ say "mystack is empty" cr cr ]
say "you are in a maze of twisty little passages, all alike"</syntaxhighlight>
 
{{out}}
 
<pre>mystack is empty
 
23 is on the top of mystack
 
mystack is on the top of mystack
 
mystack has been removed from mystack
 
23 has been removed from mystack
 
mystack is empty
 
you are in a maze of twisty little passages, all alike</pre>
 
=={{header|R}}==
{{libheader|proto}}
See [[FIFO]] for functional and object oriented implementations of a First-In-First-Out object, with similar code.
<langsyntaxhighlight Rlang="r">library(proto)
 
stack <- proto(expr = {
Line 942 ⟶ 5,733:
stack$pop()
# Error in get("pop", env = stack, inherits = TRUE)(stack, ...) :
# can't pop from an empty list</syntaxhighlight>
</lang>
 
=={{header|RavenRacket}}==
 
Quick functional version:
Use built in ''stack'' type:
 
<syntaxhighlight lang="racket">
new stack as s
#lang racket
1 s push
(define stack '())
s pop
(define (push x stack) (cons x stack))
(define (pop stack) (values (car stack) (cdr stack)))
(define (empty? stack) (null? stack))
</syntaxhighlight>
 
And a destructive object:
 
<syntaxhighlight lang="racket">
(struct stack ([items #:auto]) #:mutable #:auto-value '())
(define (push! x stack)
(set-stack-items! stack (cons x (stack-items stack))))
(define (pop! stack)
(begin0 (car (stack-items stack))
(set-stack-items! stack (cdr (stack-items stack)))))
(define (empty? stack)
(null? (stack-items stack)))
</syntaxhighlight>
 
=={{header|Raku}}==
(formerly Perl 6)
 
Raku still has the stack functions from Perl 5, but now they also can be accessed by object notation:
<syntaxhighlight lang="raku" line>my @stack; # just a array
@stack.push($elem); # add $elem to the end of @stack
$elem = @stack.pop; # get the last element back
@stack.elems == 0 # true, because the stack is empty
not @stack # also true because @stack is false</syntaxhighlight>
 
=={{header|Raven}}==
Use built in ''stack'' type:
<syntaxhighlight lang="raven">new stack as s
1 s push
s pop</syntaxhighlight>
Word ''empty'' is also built in:
<syntaxhighlight lang="raven">s empty if 'stack is empty' print</syntaxhighlight>
 
=={{header|REBOL}}==
s empty if 'stack is empty' print
<syntaxhighlight lang="rebol">REBOL [
Title: "Stack"
URL: http://rosettacode.org/wiki/Stack
]
 
stack: make object! [
=={{header|Ruby}}==
data: copy []
 
push: func [x][append data x]
Using an Array:
pop: func [/local x][x: last data remove back tail data x]
<lang ruby>
empty: does [empty? data]
stack = []
 
peek: does [last data]
]
 
; Teeny Tiny Test Suite
assert: func [code][print [either do code [" ok"]["FAIL"] mold code]]
 
print "Simple integers:"
s: make stack [] s/push 1 s/push 2 ; Initialize.
 
assert [2 = s/peek]
assert [2 = s/pop]
assert [1 = s/pop]
assert [s/empty]
 
print [lf "Symbolic data on stack:"]
v: make stack [data: [this is a test]] ; Initialize on instance.
 
assert ['test = v/peek]
assert ['test = v/pop]
assert ['a = v/pop]
assert [not v/empty]</syntaxhighlight>
Sample run:
<pre>Simple integers:
ok [2 = s/peek]
ok [2 = s/pop]
ok [1 = s/pop]
ok [s/empty]
 
Symbolic data on stack:
ok ['test = v/peek]
ok ['test = v/pop]
ok ['a = v/pop]
ok [not v/empty]
</pre>
 
=={{header|Retro}}==
<syntaxhighlight lang="retro">: stack ( n"- ) create 0 , allot ;
: push ( na- ) dup ++ dup @ + ! ;
: pop ( a-n ) dup @ over -- + @ ;
: top ( a-n ) dup @ + @ ;
: empty? ( a-f ) @ 0 = ;
 
10 stack st
 
1 st push
2 st push
3 st push
st empty? putn
st top putn
st pop putn st pop putn st pop putn
st empty? putn</syntaxhighlight>
 
=={{header|REXX}}==
===version 1===
<syntaxhighlight lang="rexx">y=123 /*define a REXX variable, value is 123 */
push y /*pushes 123 onto the stack. */
pull g /*pops last value stacked & removes it. */
q=empty() /*invokes the EMPTY subroutine (below)*/
exit /*stick a fork in it, we're done. */
 
empty: return queued() /*subroutine returns # of stacked items.*/</syntaxhighlight>
 
===version 2===
<syntaxhighlight lang="rexx">/* REXX ***************************************************************
* supports push, pull, and peek
* 11.08.2013 Walter Pachl
**********************************************************************/
stk.=0
Call push 123
Say empty()
say peek()
say pull()
Say empty()
say peek()
say push(456)
say peek()
Exit
 
push: Procedure Expose stk.
Parse Arg v
z=stk.0+1
stk.z=v
stk.0=z
Return v
 
peek: Procedure Expose stk.
If stk.0=0 Then
Return 'stack is empty'
Else Do
z=stk.0
Return stk.z
End
 
pull: Procedure Expose stk.
If stk.0=0 Then
Return 'stack is empty'
Else Do
z=stk.0
res=stk.z
stk.0=stk.0-1
Return res
End
 
empty: Procedure Expose stk.
Return stk.0=0</syntaxhighlight>
{{out}}
<pre>
0
123
123
1
stack is empty
456
456
</pre>
 
=={{header|Ring}}==
<syntaxhighlight lang="ring">
# Project : Stack
 
load "stdlib.ring"
ostack = new stack
for n = 5 to 7
see "Push: " + n + nl
ostack.push(n)
next
see "Pop:" + ostack.pop() + nl
see "Push: " + "8" + nl
ostack.push(8)
while len(ostack) > 0
see "Pop:" + ostack.pop() + nl
end
if len(ostack) = 0
see "Pop: stack is empty" + nl
ok
</syntaxhighlight>
Output:
<pre>
Push: 5
Push: 6
Push: 7
Pop:7
Push: 8
Pop:8
Pop:6
Pop:5
Pop: stack is empty
</pre>
 
=={{header|RPL}}==
The RPL interpreter is based on a stack, with which the user interacts.
*the push operation is performed by the <code>DUP</code> instruction
*the pop operation by <code>DROP</code>
*<code>DEPTH</code> provides the stack size. To test the emptiness of the stack, the following program can be created as an user-defined instruction:
≪ DEPTH NOT ≫ 'EMPTY?' STO
=={{header|Ruby}}==
Using an Array, there are already methods Array#push, Array#pop and Array#empty?.
<syntaxhighlight lang="ruby">stack = []
stack.push(value) # pushing
value = stack.pop # popping
stack.empty? # is empty?</syntaxhighlight>
If you need to expose your stack to the world, you may want to create a simpler wrapper. Here is a wrapper class ''Stack'' that wraps ''Array'' but only exposes stack methods.
</lang>
<syntaxhighlight lang="ruby">require 'forwardable'
 
# A stack contains elements in last-in, first-out order.
If you need to expose your stack to the world, you may want to create a simpler wrapper:
# Stack#push adds new elements to the top of the stack;
# Stack#pop removes elements from the top.
class Stack
extend Forwardable
# Creates a Stack containing _objects_.
def self.[](*objects)
new.push(*objects)
end
# Creates an empty Stack.
def initialize
@ary = []
end
# Duplicates a Stack.
def initialize_copy(obj)
super
@ary = @ary.dup
end
# Adds each object to the top of this Stack. Returns self.
def push(*objects)
@ary.push(*objects)
self
end
alias << push
##
# :method: pop
# :call-seq:
# pop -> obj or nil
# pop(n) -> ary
#
# Removes an element from the top of this Stack, and returns it.
# Returns nil if the Stack is empty.
#
# If passing a number _n_, removes the top _n_ elements, and returns
# an Array of them. If this Stack contains fewer than _n_ elements,
# returns them all. If this Stack is empty, returns an empty Array.
def_delegator :@ary, :pop
##
# :method: top
# :call-seq:
# top -> obj or nil
# top(n) -> ary
# Returns the topmost element without modifying the stack.
def_delegator :@ary, :last, :top
##
# :method: empty?
# Returns true if this Stack contains no elements.
def_delegator :@ary, :empty?
##
# :method: size
# Returns the number of elements in this Stack.
def_delegator :@ary, :size
alias length size
# Converts this Stack to a String.
def to_s
"#{self.class}#{@ary.inspect}"
end
alias inspect to_s
end</syntaxhighlight>
 
<syntaxhighlight lang="ruby">p s = Stack.new # => Stack[]
p s.empty? # => true
p s.size # => 0
p s.top # => nil
p s.pop # => nil
p s.pop(1) # => []
p s.push(1) # => Stack[1]
p s.push(2, 3) # => Stack[1, 2, 3]
p s.top # => 3
p s.top(2) # => [2, 3]
p s # => Stack[1, 2, 3]
p s.size # => 3
p s.pop # => 3
p s.pop(1) # => [2]
p s.empty? # => false
 
p s = Stack[:a, :b, :c] # => Stack[:a, :b, :c]
p s << :d # => Stack[:a, :b, :c, :d]
p s.pop # => :d</syntaxhighlight>
 
Just meeting the requirements of a push, pop and empty method:
<syntaxhighlight lang="ruby">require 'forwardable'
 
<lang ruby>
class Stack
extend Forwardable
def initialize
 
@items = []
def initialize
end
def@stack push(item)= []
end
@items.push(item)
 
end
def_delegators :@stack, :push, :pop, :empty?
def pop
@items.pop
end
def empty?
@items.empty?
end
end
</syntaxhighlight>
</lang>
(push takes multiple arguments; pop takes an optional argument which specifies how many to pop)
 
=={{header|Run BASIC}}==
<syntaxhighlight lang="runbasic">dim stack$(10) ' stack of ten
global stack$
global stackEnd
 
for i = 1 to 5 ' push 5 values to the stack
a$ = push$(chr$(i + 64))
print "Pushed ";chr$(i + 64);" stack has ";stackEnd
next i
 
print "Pop Value:";pop$();" stack has ";stackEnd ' pop last in
print "Pop Value:";pop$();" stack has ";stackEnd ' pop last in
 
e$ = mt$() ' MT the stack
print "Empty stack. stack has ";stackEnd
 
' ------ PUSH the stack
FUNCTION push$(val$)
stackEnd = stackEnd + 1 ' if more than 10 then lose the oldest
if stackEnd > 10 then
for i = 0 to 9
stack$(i) = stack$(i+1)
next i
stackEnd = 10
end if
stack$(stackEnd) = val$
END FUNCTION
 
' ------ POP the stack -----
FUNCTION pop$()
if stackEnd = 0 then
pop$ = "Stack is MT"
else
pop$ = stack$(stackEnd) ' pop last in
stackEnd = max(stackEnd - 1,0)
end if
END FUNCTION
 
' ------ MT the stack ------
FUNCTION mt$()
stackEnd = 0
END FUNCTION</syntaxhighlight>
{{out}}
<pre>Pushed A stack has 1
Pushed B stack has 2
Pushed C stack has 3
Pushed D stack has 4
Pushed E stack has 5
Pop Value:E stack has 4
Pop Value:D stack has 3
Empty stack. stack has 0</pre>
 
=={{header|Rust}}==
===Using the standard library===
 
One could just use a vector (<code>Vec<T></code>) which is part of the standard library
 
<syntaxhighlight lang="rust">fn main() {
let mut stack = Vec::new();
stack.push("Element1");
stack.push("Element2");
stack.push("Element3");
 
assert_eq!(Some(&"Element3"), stack.last());
assert_eq!(Some("Element3"), stack.pop());
assert_eq!(Some("Element2"), stack.pop());
assert_eq!(Some("Element1"), stack.pop());
assert_eq!(None, stack.pop());
}</syntaxhighlight>
 
===Simple implementation===
Simply uses a singly-linked list.
<syntaxhighlight lang="rust">type Link<T> = Option<Box<Frame<T>>>;
 
pub struct Stack<T> {
head: Link<T>,
}
struct Frame<T> {
elem: T,
next: Link<T>,
}
 
/// Iterate by value (consumes list)
pub struct IntoIter<T>(Stack<T>);
impl<T> Iterator for IntoIter<T> {
type Item = T;
fn next(&mut self) -> Option<Self::Item> {
self.0.pop()
}
}
 
/// Iterate by immutable reference
pub struct Iter<'a, T: 'a> {
next: Option<&'a Frame<T>>,
}
impl<'a, T> Iterator for Iter<'a, T> { // Iterate by immutable reference
type Item = &'a T;
fn next(&mut self) -> Option<Self::Item> {
self.next.take().map(|frame| {
self.next = frame.next.as_ref().map(|frame| &**frame);
&frame.elem
})
}
}
 
/// Iterate by mutable reference
pub struct IterMut<'a, T: 'a> {
next: Option<&'a mut Frame<T>>,
}
impl<'a, T> Iterator for IterMut<'a, T> {
type Item = &'a mut T;
fn next(&mut self) -> Option<Self::Item> {
self.next.take().map(|frame| {
self.next = frame.next.as_mut().map(|frame| &mut **frame);
&mut frame.elem
})
}
}
 
 
impl<T> Stack<T> {
/// Return new, empty stack
pub fn new() -> Self {
Stack { head: None }
}
 
/// Add element to top of the stack
pub fn push(&mut self, elem: T) {
let new_frame = Box::new(Frame {
elem: elem,
next: self.head.take(),
});
self.head = Some(new_frame);
}
 
/// Remove element from top of stack, returning the value
pub fn pop(&mut self) -> Option<T> {
self.head.take().map(|frame| {
let frame = *frame;
self.head = frame.next;
frame.elem
})
}
 
/// Get immutable reference to top element of the stack
pub fn peek(&self) -> Option<&T> {
self.head.as_ref().map(|frame| &frame.elem)
}
 
/// Get mutable reference to top element on the stack
pub fn peek_mut(&mut self) -> Option<&mut T> {
self.head.as_mut().map(|frame| &mut frame.elem)
}
 
/// Iterate over stack elements by value
pub fn into_iter(self) -> IntoIter<T> {
IntoIter(self)
}
 
/// Iterate over stack elements by immutable reference
pub fn iter<'a>(&'a self) -> Iter<'a,T> {
Iter { next: self.head.as_ref().map(|frame| &**frame) }
}
 
/// Iterate over stack elements by mutable reference
pub fn iter_mut(&mut self) -> IterMut<T> {
IterMut { next: self.head.as_mut().map(|frame| &mut **frame) }
}
}
 
// The Drop trait tells the compiler how to free an object after it goes out of scope.
// By default, the compiler would do this recursively which *could* blow the stack for
// extraordinarily long lists. This simply tells it to do it iteratively.
impl<T> Drop for Stack<T> {
fn drop(&mut self) {
let mut cur_link = self.head.take();
while let Some(mut boxed_frame) = cur_link {
cur_link = boxed_frame.next.take();
}
}
}</syntaxhighlight>
 
=={{header|Sather}}==
This one uses a builtin linked list to keep the values pushed onto the stack.
<syntaxhighlight lang="sather">class STACK{T} is
private attr stack :LLIST{T};
 
create:SAME is
res ::= new;
res.stack := #LLIST{T};
return res;
end;
 
push(elt: T) is
stack.insert_front(elt);
end;
 
pop: T is
if ~stack.is_empty then
stack.rewind;
r ::= stack.current;
stack.delete;
return r;
else
raise "stack empty!\n";
end;
end;
 
top: T is
stack.rewind;
return stack.current;
end;
 
is_empty: BOOL is
return stack.is_empty;
end;
end;</syntaxhighlight>
 
<syntaxhighlight lang="sather">class MAIN is
main is
s ::= #STACK{INT};
#OUT + "push values...\n";
s.push(3);
s.push(2);
s.push(1);
s.push(0);
#OUT + "retrieving them...\n";
loop
#OUT + s.pop + "\n";
until!(s.is_empty); end;
end;
end;</syntaxhighlight>
Sather library has the abstract class <code>$STACK{T}</code>, but using this forces us to implement other methods too.
 
=={{header|Scala}}==
The Do it yourself approach:
<syntaxhighlight lang="scala">class Stack[T] {
private var items = List[T]()
 
def isEmpty = items.isEmpty
 
def peek = items match {
case List() => error("Stack empty")
case head :: rest => head
}
 
def pop = items match {
case List() => error("Stack empty")
case head :: rest => items = rest; head
}
 
def push(value: T) = items = value +: items
}</syntaxhighlight>
Or use the standard Scala library.
Slightly modified to meet to requirements of this task.
<syntaxhighlight lang="scala">import collection.mutable.{ Stack => Stak }
 
class Stack[T] extends Stak[T] {
override def pop: T = {
if (this.length == 0) error("Can't Pop from an empty Stack.")
else super.pop
}
def peek: T = this.head
}</syntaxhighlight>A test could be:<syntaxhighlight lang="scala">object StackTest extends App {
 
val stack = new Stack[String]
 
stack.push("Peter Pan")
stack.push("Suske & Wiske", "Alice in Wonderland")
 
assert(stack.peek == "Alice in Wonderland")
assert(stack.pop() == "Alice in Wonderland")
assert(stack.pop() == "Suske & Wiske")
assert(stack.pop() == "Peter Pan")
println("Completed without errors")
}</syntaxhighlight>
 
=={{header|Scheme}}==
This version uses primitive message passing.
<syntaxhighlight lang="scheme">(define (make-stack)
<lang scheme>
(define (make-stack)
(let ((st '()))
(lambda (message . args)
Line 1,003 ⟶ 6,353:
(set! st (cdr st))
result)))
(else 'badmsg)))))</syntaxhighlight>
 
</lang>
=={{header|Seed7}}==
<syntaxhighlight lang="seed7">$ include "seed7_05.s7i";
 
const func type: stack (in type: baseType) is func
result
var type: stackType is void;
begin
stackType := array baseType;
 
const proc: push (inout stackType: aStack, in baseType: top) is func
begin
aStack := [] (top) & aStack;
end func;
 
const func baseType: pop (inout stackType: aStack) is func
result
var baseType: top is baseType.value;
begin
if length(aStack) = 0 then
raise RANGE_ERROR;
else
top := aStack[1];
aStack := aStack[2 ..];
end if;
end func;
 
const func boolean: empty (in stackType: aStack) is
return length(aStack) = 0;
end func;
 
const type: intStack is stack(integer);
 
const proc: main is func
local
var intStack: s is intStack.value;
begin
push(s, 10);
push(s, 20);
writeln(pop(s) = 20);
writeln(pop(s) = 10);
writeln(empty(s));
end func;</syntaxhighlight>
 
=={{header|SenseTalk}}==
<syntaxhighlight lang="sensetalk">put () into stack
repeat with each item of 1 .. 10
push it into stack
end repeat
 
repeat while stack is not empty
pop stack
put it
end repeat</syntaxhighlight>
 
=={{header|Sidef}}==
Using a built-in array:
<syntaxhighlight lang="ruby">var stack = [];
stack.push(42); # pushing
say stack.pop; # popping
say stack.is_empty; # is_emtpy?</syntaxhighlight>
 
Creating a Stack class:
<syntaxhighlight lang="ruby">class Stack(stack=[]) {
method pop { stack.pop };
method push(item) { stack.push(item) };
method empty { stack.is_empty };
}
 
var stack = Stack();
stack.push(42);
say stack.pop; # => 42
say stack.empty; # => true</syntaxhighlight>
 
=={{header|Slate}}==
From Slate's standard library:
<syntaxhighlight lang="slate">collections define: #Stack &parents: {ExtensibleArray}.
<lang slate>
collections define: #Stack &parents: {ExtensibleArray}.
"An abstraction over ExtensibleArray implementations to follow the stack
protocol. The convention is that the Sequence indices run least-to-greatest
Line 1,030 ⟶ 6,451:
 
s@(Stack traits) bottom
[s first].</syntaxhighlight>
 
</lang>
=={{header|Smalltalk}}==
Smalltalk has a built-in Stack class, instances of which you can send messages:
<syntaxhighlight lang="smalltalk">
s := Stack new.
s push: 1.
s push: 2.
s push: 3.
s pop.
s top. "2"
</syntaxhighlight>
 
=={{header|Standard ML}}==
 
The signature for a module supplying a stack interface, with a couple added functions.
 
<syntaxhighlight lang="sml">signature STACK =
sig
type 'a stack
exception EmptyStack
 
val empty : 'a stack
val isEmpty : 'a stack -> bool
 
val push : ('a * 'a stack) -> 'a stack
val pop : 'a stack -> 'a stack
val top : 'a stack -> 'a
val popTop : 'a stack -> 'a stack * 'a
 
val map : ('a -> 'b) -> 'a stack -> 'b stack
val app : ('a -> unit) -> 'a stack -> unit
end</syntaxhighlight>
 
An implementation of the <code>STACK</code> signature, using immutable lists.
 
<syntaxhighlight lang="sml">structure Stack :> STACK =
struct
type 'a stack = 'a list
exception EmptyStack
 
val empty = []
 
fun isEmpty st = null st
 
fun push (x, st) = x::st
 
fun pop [] = raise EmptyStack
| pop (x::st) = st
 
fun top [] = raise EmptyStack
| top (x::st) = x
 
fun popTop st = (pop st, top st)
 
fun map f st = List.map f st
fun app f st = List.app f st
end</syntaxhighlight>
 
=={{header|Stata}}==
See [[Singly-linked list/Element definition#Stata]].
 
=={{header|Swift}}==
Generic stack.
<syntaxhighlight lang="swift">struct Stack<T> {
var items = [T]()
var empty:Bool {
return items.count == 0
}
func peek() -> T {
return items[items.count - 1]
}
mutating func pop() -> T {
return items.removeLast()
}
mutating func push(obj:T) {
items.append(obj)
}
}
 
var stack = Stack<Int>()
stack.push(1)
stack.push(2)
println(stack.pop())
println(stack.peek())
stack.pop()
println(stack.empty)</syntaxhighlight>
{{out}}
<pre>
2
1
true
</pre>
 
=={{header|Tailspin}}==
<syntaxhighlight lang="tailspin">
processor Stack
@: $;
 
sink push
..|@Stack: $;
end push
 
source peek
$@Stack(last) !
end peek
 
source pop
^@Stack(last) !
end pop
 
source empty
$@Stack::length -> #
<=0> 1 !
<> 0 !
end empty
end Stack
 
def myStack: [1] -> Stack;
 
2 -> !myStack::push
 
'$myStack::empty; $myStack::pop;
' -> !OUT::write
'$myStack::empty; $myStack::pop;
' -> !OUT::write
'$myStack::empty;
' -> !OUT::write
 
3 -> !myStack::push
'$myStack::empty; $myStack::peek;
' -> !OUT::write
'$myStack::empty; $myStack::pop;
' -> !OUT::write
'$myStack::empty;' -> !OUT::write
</syntaxhighlight>
{{out}}
<pre>
0 2
0 1
1
0 3
0 3
1
</pre>
 
=={{header|Tcl}}==
Here's a simple implementation using a list:
<langsyntaxhighlight lang="tcl">proc push {stackvar value} {
upvar 1 $stackvar stack
lappend stack $value
Line 1,065 ⟶ 6,632:
peek S ;# ==> bar
pop S ;# ==> bar
peek S ;# ==> foo</langsyntaxhighlight>
{{tcllib|struct::stack}}
 
<syntaxhighlight lang="tcl">package require struct::stack
There is a package in {{libheader|tcllib}}called <code>struct::stack</code> that presents an object interface:
 
<lang tcl>package require struct::stack
struct::stack S
S size ;# ==> 0
Line 1,078 ⟶ 6,643:
S peek ;# ==> d
S pop 4 ;# ==> d c b a
S size ;# ==> 0</langsyntaxhighlight>
 
=={{header|Uiua}}==
 
<syntaxhighlight lang="Uiua">
[3] # Since UIUA is a stack language, everything is pushed on the stack
x ← # stores the top of the stack into the variable x
? # ? checks the stack, it is now empty
 
</syntaxhighlight>
 
 
 
=={{header|UnixPipes}}==
<syntaxhighlight lang="bash">init() { if [ -e stack ]; then rm stack; fi } # force pop to blow up if empty
Uses moreutils
initpush() { echo $1 >> stack); }
pop() {
push() {echo $1 >> stack}
pop() {tail -1 stack; (cat stack |head -n -1)|sponge stack}
empty()x=`head {cat-n -1 stack | wc -l}c`
if [ $x -eq '0' ]; then rm stack; else
Usage:
truncate -s `head -n -1 stack | wc -c` stack
push me; push you; push us; push them
fi
pop;pop;pop;pop
}
them
empty() { head -n -1 stack |wc -l; }
us
stack_top() { tail -1 stack; }</syntaxhighlight>
you
test it:
me
<syntaxhighlight lang="bash">% push me; push you; push us; push them
% pop;pop;pop;pop
them
us
you
me</syntaxhighlight>
 
=={{header|UNIX Shell}}==
{{works with|Bourne Again SHell}}
{{works with|Zsh}}
{{works with|Korn Shell}}
 
Here's a simple single-stack solution:
<syntaxhighlight lang="sh">init() {
if [[ -n $KSH_VERSION ]]; then
set -A stack
else
stack=(); # this sets stack to '()' in ksh
fi
}
 
push() {
stack=("$1" "${stack[@]}")
}
 
stack_top() {
# this approach sidesteps zsh indexing difference
set -- "${stack[@]}"
printf '%s\n' "$1"
}
 
pop() {
stack_top
stack=("${stack[@]:1}")
}
 
empty() {
(( ${#stack[@]} == 0 ))
}
 
# Demo
push fred; push wilma; push betty; push barney
printf 'peek(stack)==%s\n' "$(stack_top)"
while ! empty; do
pop
done</syntaxhighlight>
 
{{Out}}
<pre>peek(stack)==barney
barney
betty
wilma
fred</pre>
 
You can generalize it to multiple stacks with some judicious use of the twin evils of pass-by-name and <tt>eval</tt>:
 
<syntaxhighlight lang="sh">init_stack() {
if [[ -n $KSH_VERSION ]]; then
eval 'set -A '"$1"
else
eval "$1=()"
fi
}
 
push() {
eval "$1"'=("$2" "${'"$1"'[@]}")'
}
 
stack_top() {
eval 'set -- "${'"$1"'[@]}"';
printf '%s\n' "$1"
}
 
pop() {
stack_top "$1";
eval "$1"'=("${'"$1"'[@]:1}")'
}
 
empty() {
eval '(( ${#'"$1"'[@]} == 0 ))'
}
 
init_stack mystack
push mystack fred; push mystack wilma; push mystack betty; push mystack barney
printf 'peek(mystack)==%s\n' "$(stack_top mystack)"
while ! empty mystack; do
pop mystack
done</syntaxhighlight>
 
{{Out}}
<pre>peek(mystack)==barney
barney
betty
wilma
fred</pre>
 
=={{header|VBA}}==
Define a class Stack in a class module with that name.
<syntaxhighlight lang="vb">'Simple Stack class
 
'uses a dynamic array of Variants to stack the values
'has read-only property "Size"
'and methods "Push", "Pop", "IsEmpty"
 
Private myStack()
Private myStackHeight As Integer
 
'method Push
Public Function Push(aValue)
'increase stack height
myStackHeight = myStackHeight + 1
ReDim Preserve myStack(myStackHeight)
myStack(myStackHeight) = aValue
End Function
 
'method Pop
Public Function Pop()
'check for nonempty stack
If myStackHeight > 0 Then
Pop = myStack(myStackHeight)
myStackHeight = myStackHeight - 1
Else
MsgBox "Pop: stack is empty!"
End If
End Function
 
'method IsEmpty
Public Function IsEmpty() As Boolean
IsEmpty = (myStackHeight = 0)
End Function
 
'property Size
Property Get Size() As Integer
Size = myStackHeight
End Property</syntaxhighlight>
Usage example:
<syntaxhighlight lang="vb">'stack test
Public Sub stacktest()
Dim aStack As New Stack
With aStack
'push and pop some value
.Push 45
.Push 123.45
.Pop
.Push "a string"
.Push "another string"
.Pop
.Push Cos(0.75)
Debug.Print "stack size is "; .Size
While Not .IsEmpty
Debug.Print "pop: "; .Pop
Wend
Debug.Print "stack size is "; .Size
'try to continue popping
.Pop
End With
End Sub</syntaxhighlight>
{{out}}
<pre>
stacktest
stack size is 3
pop: 0,731688868873821
pop: a string
pop: 45
stack size is 0
</pre>
(after wich a message box will pop up)
 
=={{header|VBScript}}==
===Stack class===
<syntaxhighlight lang="vb">class stack
dim tos
dim stack()
dim stacksize
private sub class_initialize
stacksize = 100
redim stack( stacksize )
tos = 0
end sub
 
public sub push( x )
stack(tos) = x
tos = tos + 1
end sub
public property get stackempty
stackempty = ( tos = 0 )
end property
public property get stackfull
stackfull = ( tos > stacksize )
end property
public property get stackroom
stackroom = stacksize - tos
end property
public function pop()
pop = stack( tos - 1 )
tos = tos - 1
end function
 
public sub resizestack( n )
redim preserve stack( n )
stacksize = n
if tos > stacksize then
tos = stacksize
end if
end sub
end class
 
dim s
set s = new stack
s.resizestack 10
wscript.echo s.stackempty
dim i
for i = 1 to 10
s.push rnd
wscript.echo s.stackroom
if s.stackroom = 0 then exit for
next
for i = 1 to 10
wscript.echo s.pop
if s.stackempty then exit for
next</syntaxhighlight>
{{out}} (changes every time)
<pre>-1
9
8
7
6
5
4
3
2
1
0
0.7090379
0.81449
0.7607236
1.401764E-02
0.7747401
0.301948
0.2895625
0.5795186
0.533424
0.7055475</pre>
===Using an ArrayList.===
<syntaxhighlight lang="vb">' Stack Definition - VBScript
Option Explicit
 
Dim stack, i, x
Set stack = CreateObject("System.Collections.ArrayList")
If Not empty_(stack) Then Wscript.Echo stack.Count
push stack, "Banana"
push stack, "Apple"
push stack, "Pear"
push stack, "Strawberry"
Wscript.Echo "Count=" & stack.Count ' --> Count=4
Wscript.Echo pop(stack) & " - Count=" & stack.Count ' --> Strawberry - Count=3
Wscript.Echo "Tail=" & stack.Item(0) ' --> Tail=Banana
Wscript.Echo "Head=" & stack.Item(stack.Count-1) ' --> Head=Pear
Wscript.Echo stack.IndexOf("Apple", 0) ' --> 1
For i=1 To stack.Count
Wscript.Echo join(stack.ToArray(), ", ")
x = pop(stack)
Next 'i
 
Sub push(s, what)
s.Add what
End Sub 'push
Function pop(s)
Dim what
If s.Count > 0 Then
what = s(s.Count-1)
s.RemoveAt s.Count-1
Else
what = ""
End If
pop = what
End Function 'pop
Function empty_(s)
empty_ = s.Count = 0
End Function 'empty_ </syntaxhighlight>
{{out}}
<pre>
Count=4
Strawberry - Count=3
Tail=Banana
Head=Pear
1
Banana, Apple, Pear
Banana, Apple
Banana
</pre>
 
=={{header|V (Vlang)}}==
<syntaxhighlight lang="go">const (
max_depth = 256
)
 
struct Stack {
mut:
data []f32 = []f32{len: max_depth}
depth int
}
 
fn (mut s Stack) push(v f32) {
if s.depth >= max_depth {
return
}
println('Push: ${v:3.2f}')
s.data[s.depth] = v
s.depth++
}
 
fn (mut s Stack) pop() ?f32 {
if s.depth > 0 {
s.depth--
result := s.data[s.depth]
println('Pop: top of stack was ${result:3.2f}')
return result
}
return error('Stack Underflow!!')
}
 
fn (s Stack) peek() ?f32 {
if s.depth > 0 {
result := s.data[s.depth - 1]
println('Peek: top of stack is ${result:3.2f}')
return result
}
return error('Out of Bounds...')
}
 
fn (s Stack) empty() bool {
return s.depth == 0
}
 
fn main() {
mut stack := Stack{}
println('Stack is empty? ' + if stack.empty() { 'Yes' } else { 'No' })
stack.push(5.0)
stack.push(4.2)
println('Stack is empty? ' + if stack.empty() { 'Yes' } else { 'No' })
stack.peek() or { return }
stack.pop() or { return }
stack.pop() or { return }
}
</syntaxhighlight>
{{out}}
<pre>Stack is empty? Yes
Push: 5.00
Push: 4.20
Stack is empty? No
Peek: top of stack is 4.20
Pop: top of stack was 4.20
Pop: top of stack was 5.00
</pre>
 
=={{header|Wart}}==
 
Stacks as user-defined objects backed by a list.
 
<syntaxhighlight lang="wart">def (stack)
(tag 'stack nil)
 
mac (push! x s) :qcase `(isa stack ,s)
`(push! ,x (rep ,s))
 
mac (pop! s) :qcase `(isa stack ,s)
`(pop! (rep ,s))
 
def (empty? s) :case (isa stack s)
(empty? rep.s)</syntaxhighlight>
 
Example usage:
 
<pre>s <- (stack)
=> (object stack nil)
push! 3 s
=> (object stack (3))
push! 4 s
=> (object stack (4 3))
push! 5 s
=> (object stack (5 4 3))
pop! s
=> 5
(empty? s)
=> nil
pop! s
=> 4
pop! s
=> 3
(empty? s)
=> 1 # true</pre>
 
=={{header|Wren}}==
{{libheader|Wren-seq}}
This uses the Stack class in the above module.
<syntaxhighlight lang="wren">import "./seq" for Stack
 
var s = Stack.new()
s.push(1)
s.push(2)
System.print("Stack contains %(s.toList)")
System.print("Number of elements in stack = %(s.count)")
var item = s.pop()
System.print("'%(item)' popped from the stack")
System.print("Last element is now %(s.peek())")
s.clear()
System.print("Stack cleared")
System.print("Is stack now empty? %((s.isEmpty) ? "yes" : "no")")</syntaxhighlight>
 
{{out}}
<pre>
Stack contains [1, 2]
Number of elements in stack = 2
'2' popped from the stack
Last element is now 1
Stack cleared
Is stack now empty? yes
</pre>
 
=={{header|X86 Assembly}}==
<syntaxhighlight lang="x86asm">
; x86_64 linux nasm
 
struc Stack
maxSize: resb 8
currentSize: resb 8
contents:
endStruc
 
section .data
 
soError: db "Stack Overflow Exception", 10
seError: db "Stack Empty Error", 10
 
 
section .text
 
createStack:
; IN: max number of elements (rdi)
; OUT: pointer to new stack (rax)
push rdi
xor rdx, rdx
mov rbx, 8
mul rbx
mov rcx, rax
mov rax, 12
mov rdi, 0
syscall
push rax
mov rdi, rax
add rdi, rcx
mov rax, 12
syscall
pop rax
pop rbx
mov qword [rax + maxSize], rbx
mov qword [rax + currentSize], 0
ret
 
 
push:
; IN: stack to operate on (stack argument), element to push (rdi)
; OUT: void
mov rax, qword [rsp + 8]
mov rbx, qword [rax + currentSize]
cmp rbx, qword [rax + maxSize]
je stackOverflow
lea rsi, [rax + contents + 8*rbx]
mov qword [rsi], rdi
add qword [rax + currentSize], 1
ret
 
 
pop:
; pop
; IN: stack to operate on (stack argument)
; OUT: element from stack top
mov rax, qword [rsp + 8]
mov rbx, qword [rax + currentSize]
cmp rbx, 0
je stackEmpty
sub rbx, 1
lea rsi, [rax + contents + 8*rbx]
mov qword [rax + currentSize], rbx
mov rax, qword [rsi]
ret
 
 
; stack operation exceptions
stackOverflow:
mov rsi, soError
mov rdx, 25
jmp errExit
stackEmpty:
mov rsi, seError
mov rdx, 18
errExit:
mov rax, 1
mov rdi, 1
syscall
mov rax, 60
mov rdi, 1
syscall
</syntaxhighlight>
 
=={{header|XLISP}}==
This is a fairly straightforward implementation, representing a stack as a linked list inside an object.
<syntaxhighlight lang="lisp">(define-class stack
(instance-variables vals))
 
(define-method (stack 'initialize)
(setq vals '())
self)
 
(define-method (stack 'push x)
(setq vals (cons x vals)))
 
(define-method (stack 'pop)
(define tos (car vals))
(setq vals (cdr vals))
tos)
 
(define-method (stack 'emptyp)
(null vals))</syntaxhighlight>
A sample REPL session:
<syntaxhighlight lang="lisp">; Loading 'stack.lsp'
[1] (define st (stack 'new))
 
ST
[2] (st 'push 1)
 
(1)
[3] (st 'push 2)
 
(2 1)
[4] (st 'emptyp)
 
()
[5] (st 'pop)
 
2
[6] (st 'pop)
 
1
[7] (st 'emptyp)
 
#T
[8] </syntaxhighlight>
 
=={{header|XPL0}}==
<syntaxhighlight lang="xpl0">include c:\cxpl\codes; \intrinsic 'code' declarations
int Stack(100), SP;
 
proc Push(I); \Push an integer onto the Stack
int I;
[SP:= SP+1;
Stack(SP):= I;
]; \Push
 
func Pop; \Pop an integer from the Stack
int I;
[I:= Stack(SP);
SP:= SP-1;
return I;
]; \Pop
 
func Empty; \Return 'true' if Stack is empty
return SP<0;
 
func Top; \Return the integer at top of Stack
return Stack(SP);
 
int I;
[SP:= -1; \initialize stack pointer
for I:= 0 to 10 do Push(I*I);
IntOut(0, Top); CrLf(0);
while not Empty do [IntOut(0, Pop); ChOut(0, ^ )];
CrLf(0);
]</syntaxhighlight>
 
{{out}}
<pre>
100
100 81 64 49 36 25 16 9 4 1 0
</pre>
 
=={{header|Yabasic}}==
<syntaxhighlight lang="yabasic">limit = 1000
dim stack(limit)
 
top = 0
 
sub push(n)
if top < limit then
top = top + 1 : stack(top) = n
else
print "stack full - ";
end if
end sub
 
sub pop()
if top then
top = top - 1 : return stack(top + 1)
else
print "stack empty - ";
end if
end sub
 
sub empty()
return not top
end sub
 
// ======== test ========
 
for n = 3 to 5
print "Push ", n : push(n)
next
 
print "Pop ", pop()
 
print "Push ", 6 : push(6)
 
while(not empty())
print "Pop ", pop()
wend
 
print "Pop ", pop()
</syntaxhighlight>
=={{header|Z80 Assembly}}==
 
The stack can be initialized by loading it directly with an immediate value. Z80-based home computers such as the Amstrad CPC and ZX Spectrum do this for you. Messing with the stack on those systems is a bad idea, since an assembly program stored on a floppy disk or cassette tape begins with the return address of BASIC on top of the stack. However, on embedded systems like the Game Boy or the Sega Master System, this step is a must, as the CPU does ''not'' have an initial stack pointer value in its vector table and thus does not guarantee the value of SP upon startup. Unlike the 6502, the Z80's stack does not have a fixed size or memory location, and is only limited by the address space of the CPU. From a practical standpoint, however, it's very unlikely you'll need more than 256 bytes.
 
<syntaxhighlight lang="z80">LD SP,&FFFF</syntaxhighlight>
 
 
Registers must be pushed in pairs. If you push/pop the accumulator, the processor flags go with it. This can make certain functions difficult to write without using a temporary variable to hold the accumulator, which doesn't allow for recursion or arbitrary nesting.
<syntaxhighlight lang="z80">push af
push bc
push de
push hl</syntaxhighlight>
 
Popping is very similar. To properly pop values, they must be popped in the reverse order they were pushed.
<syntaxhighlight lang="z80">pop hl
pop de
pop bc
pop af</syntaxhighlight>
 
The stack is empty if its value equals the original starting value of the stack pointer. This is a little difficult, since the stack doesn't necessarily start in a fixed location like it does on the 6502. There are two ways to do this:
 
<syntaxhighlight lang="z80">ld (&nnnn),SP ;&nnnn represents a memory location that the programmer will later read from to use as a
;comparison for the current stack pointer</syntaxhighlight>
 
<syntaxhighlight lang="z80">ld hl,0
add hl,sp ;the z80 doesn't allow you to load SP directly into HL, so this is the quickest way</syntaxhighlight>
 
From there it's a matter of comparing this value to the current stack pointer, which in itself is tricky since the built-in compare instruction forces you to use the accumulator as one of the operands, and works natively in terms of 8-bit values.
 
Peek can be achieved with the <code>EX (SP),HL</code> command which exchanges HL with the top item of the stack.
 
On the Game Boy, the stack can also be manually adjusted by a signed 8-bit constant. A Zilog Z80 cannot do this in a single command. The code below only works on a Game Boy or any other hardware running on a Sharp LR35902 CPU:
<syntaxhighlight lang="z80"> ADD SP,&FE ;subtract two from the stack pointer. Remember that the stack grows "down" in memory.</syntaxhighlight>
 
It should be noted that although the "heap" and the "stack" are considered separate areas of memory by the programmer, in the eyes of the CPU there is no boundary between them. The CPU doesn't care if you push enough words onto the stack so that the stack pointer is now pointing to the heap, or even ROM space. Most of the time this isn't an issue, as long as your push/pop operations are properly balanced. It's just something to look out for.
 
=={{header|zkl}}==
Lists have stack methods so this class is somewhat reduntant
<syntaxhighlight lang="zkl">class Stack{
var [const] stack=L();
fcn push(x){stack.append(x); self}
fcn pop {stack.pop()}
fcn empty {(not stack.len())}
var [proxy] isEmpty = empty;
}</syntaxhighlight>
{{out}}
<pre>
var s=Stack();
s.push(5).push("five");
s.isEmpty //-->False
s.pop() //-->"five"
</pre>
1

edit