Jump to content

Null object

From Rosetta Code
Revision as of 05:22, 13 October 2024 by Peak (talk | contribs) (JSON `null`)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Task
Null object
You are encouraged to solve this task according to the task description, using any language you may know.

Null (or nil) is the computer science concept of an undefined or unbound object. Some languages have an explicit way to access the null object, and some don't. Some languages distinguish the null object from undefined values, and some don't.


Task

Show how to access null in your language by checking to see if an object is equivalent to the null object.


This task is not about whether a variable is defined. The task is about "null"-like values in various languages, which may or may not be related to the defined-ness of variables in your language.

11l

F f([Int]? &a)
   I a != N
      a.append(1)

f(N)
[Int] arr
f(&arr)
print(arr)
Output:
[1]

6502 Assembly

Translation of: Z80 Assembly

Technically there is no such thing as a null pointer; all pointers point to something. It's a matter of what you're willing to give up. Often the null pointer is thought of as memory address 0, and on many CPUs this is the case - however this is most assuredly not the case on the 6502. Reason being, zero page RAM is limited and quite valuable, given that there are only 256 bytes of it and it is much more efficient to access than regular memory. As such, declaring $0000 to be the null pointer would be a very poor choice. Ideally, the null pointer on a 6502 should:

  • Be somewhere that isn't zero page RAM
  • Be somewhere that we cannot change at runtime (e.g. read-only memory) or doing so would cause major problems (the vector table)
  • Point to something that has no value to the programmer.


We can choose quite a few places, the easiest one I can think of is $FFFF. Although it sort of breaks our first rule, as when dereferenced as a 16-bit value, you get the value stored at $0000 as the high byte, we still can access $0000 normally anyway. Since it points to the high byte of the interrupt request vector, it's something we don't want to (or most likely can't) modify at runtime, and is of no use to us (if we really wanted the IRQ handler's address we'd dereference $FFFE instead.)

How a null pointer is implemented is very simple. You decide beforehand what your null pointer will be, and before you dereference a pointer variable, compare it to the null pointer, and if they're equal, don't dereference it. That's all there is to it.

lda pointer  ;a zero-page address that holds the low byte of a pointer variable.
CMP #$FF
BNE .continue
lda pointer+1
CMP #$FF
BNE .continue
RTS            ;return without doing anything
.continue

8th

null? if "item was null" . then

AArch64 Assembly

Works with: as version Raspberry Pi 3B version Buster 64 bits
/* ARM assembly AARCH64 Raspberry PI 3B */
/*  program nullobj64.s   */
 
/*******************************************/
/* Constantes file                         */
/*******************************************/
/* for this file see task include a file in language AArch64 assembly*/
.include "../includeConstantesARM64.inc"
/*******************************************/
/* Initialized data                        */
/*******************************************/
.data
szCarriageReturn:       .asciz "\n"
szMessResult:           .asciz "Value is null.\n"     // message result
 
qPtrObjet:              .quad 0                       // objet pointer
/*******************************************/
/* UnInitialized data                      */
/*******************************************/
.bss 
/*******************************************/
/*  code section                           */
/*******************************************/
.text
.global main 
main:                                                // entry of program
 
    ldr x0,qAdrqPtrObjet                             // load pointer address
    ldr x0,[x0]                                      // load pointer value
    cbnz x0,100f                                     // is null ?
    ldr x0,qAdrszMessResult                          // yes -> display message
    bl affichageMess
 
100:                                                 // standard end of the program
    mov x0,0                                         // return code
    mov x8,EXIT                                      // request to exit program
    svc 0                                            // perform the system call
 
qAdrszMessResult:        .quad szMessResult
qAdrszCarriageReturn:    .quad szCarriageReturn
qAdrqPtrObjet:           .quad qPtrObjet
/********************************************************/
/*        File Include fonctions                        */
/********************************************************/
/* for this file see task include a file in language AArch64 assembly */
.include "../includeARM64.inc"

Action!

TYPE Object=[
  BYTE byteData
  INT intData
  CARD cardData]

PROC IsNull(Object POINTER ptr)
  IF ptr=0 THEN
    PrintE("Object is null")
  ELSE
    PrintE("Object is not null")
  FI
RETURN

PROC Main()
  Object a
  Object POINTER ptr1=a,ptr2=0

  IsNull(ptr1)
  IsNull(ptr2)
RETURN
Output:

Screenshot from Atari 8-bit computer

Object is not null
Object is null

ActionScript

if (object == null)
    trace("object is null");

ActionScript also has an undefined value: see Undefined values#ActionScript.

Ada

with Ada.Text_Io;

if Object = null then
   Ada.Text_Io.Put_line("object is null");
end if;

ALGOL 68

In ALGOL 68 the NIL yields a name that does not refer to any value. NIL can never be naturally coerced and can only appear where the context is strong.

Works with: ALGOL 68 version Revision 1 - no extensions to language used
Works with: ALGOL 68G version Any - tested with release 1.18.0-9h.tiny
Works with: ELLA ALGOL 68 version Any (with appropriate job cards) - tested with release 1.8-8d
REF STRING no result = NIL;
STRING result := "";
 
IF no result :=: NIL THEN print(("no result :=: NIL", new line)) FI;
IF result :/=: NIL THEN print(("result :/=: NIL", new line)) FI;
 
IF no result IS NIL THEN print(("no result IS NIL", new line)) FI;
IF result ISNT NIL THEN print(("result ISNT NIL", new line)) FI;
 
COMMENT using the UNESCO/IFIP/WG2.1 ALGOL 68 character set
  result := °;
  IF REF STRING(result) :≠: ° THEN print(("result ≠ °", new line)) FI;
END COMMENT

# Note the following gotcha: #

REF STRING var := NIL;
IF var ISNT NIL THEN print(("The address of var ISNT NIL",new line)) FI;
IF var IS REF STRING(NIL) THEN print(("The address of var IS REF STRING(NIL)",new line)) FI

Output:

no result :=: NIL
result :/=: NIL
no result IS NIL
result ISNT NIL
The address of var ISNT NIL
The address of var IS REF STRING(NIL)

NIL basically is an untyped ref (pointer) that does not refer anywhere.

ALGOL 68 also has empty. This is a "constant" of size 0 and type void. c.f. Roots of a function for two different examples of usage.

  • empty as an undefined argument to a routine.
  • empty as a routine return if no result is found.

empty is typically used to refer to am empty leaf in a tree structure.

Basically:

  • ALGOL 68's empty is python's None,
  • ALGOL 68's void is python's NoneType, and
  • ALGOL 68's nil is python's hash(None)

ALGOL W

begin
    % declare a record type - will be accessed via references                %
    record R( integer f1, f2, f3 );
    % declare a reference to a R instance                                    %
    reference(R) refR;
    % assign null to the reference                                           %
    refR := null;
    % test for a null reference - will write "refR is null"                  %
    if refR = null then write( "refR is null" ) else write( "not null" );
end.

AmigaE

DEF x : PTR TO object
-> ...
IF object <> NIL
  -> ...
ENDIF

APL

APL is a vector/array-based language, so rather than a 'null pointer' or 'null value' there is the 'null vector'.

      ⍝⍝ GNU APL
      ]help 
    niladic function: Z    (Zilde)
    Zilde is the empty numeric vector (aka. 0)
Not a function but rather an alias for the empty
vector:
      ≡⍳0
1

AppleScript

Many applications will return missing value, but null is also available.

if x is missing value then
  display dialog "x is missing value"
end if

if x is null then
  display dialog "x is null"
end if

ARM Assembly

Works with: as version Raspberry Pi
/* ARM assembly Raspberry PI  */
/*  program nullobj.s   */

/* Constantes    */
.equ STDIN,  0                                        @ Linux input console
.equ STDOUT, 1                                        @ Linux output console
.equ EXIT,   1                                        @ Linux syscall
.equ READ,   3                                        @ Linux syscall
.equ WRITE,  4                                        @ Linux syscall

/* Initialized data */
.data
szCarriageReturn:       .asciz "\n"
szMessResult:           .asciz "Value is null.\n"     @ message result

iPtrObjet:		.int 0                        @ objet pointer

/* UnInitialized data */
.bss 

/*  code section */
.text
.global main 
main:                                                @ entry of program

    ldr r0,iAdriPtrObjet                             @ load pointer address
    ldr r0,[r0]                                      @ load pointer value
    cmp r0,#0                                        @ is null ?
    ldreq r0,iAdrszMessResult                        @ yes -> display message
    bleq affichageMess


100:                                                 @ standard end of the program
    mov r0, #0                                       @ return code
    pop {fp,lr}                                      @ restaur 2 registers
    mov r7, #EXIT                                    @ request to exit program
    svc 0                                            @ perform the system call

iAdrszMessResult:        .int szMessResult
iAdrszCarriageReturn:    .int szCarriageReturn
iAdriPtrObjet:           .int iPtrObjet
/******************************************************************/
/*     display text with size calculation                         */ 
/******************************************************************/
/* r0 contains the address of the message */
affichageMess:
    push {r0,r1,r2,r7,lr}                           @ save  registres
    mov r2,#0                                       @ counter length 
1:                                                  @ loop length calculation 
    ldrb r1,[r0,r2]                                 @ read octet start position + index 
    cmp r1,#0                                       @ if 0 its over 
    addne r2,r2,#1                                  @ else add 1 in the length 
    bne 1b                                          @ and loop 
                                                    @ so here r2 contains the length of the message 
    mov r1,r0                                       @ address message in r1 
    mov r0,#STDOUT                                  @ code to write to the standard output Linux 
    mov r7, #WRITE                                  @ code call system "write" 
    svc #0                                          @ call systeme 
    pop {r0,r1,r2,r7,lr}                            @ restaur des  2 registres */ 
    bx lr                                           @ return

Arturo

v: null
 
if v=null -> print "got NULL!"
Output:
got NULL!

AutoHotkey

If (object == null)
  MsgBox, object is null

AutoIt

Local $object = Null
If $object = Null Then MsgBox(0, "NULL", "Object is null")

AWK

Undefined elements correspond to an empty string; when converted to a numerical value, it evaluates to 0. In order to distinguish a undefined value from a value of 0, length(var) need to be used.

#!/usr/bin/awk -f
BEGIN { 
  b=0; 
  print "<"b,length(b)">" 
  print "<"u,length(u)">" 
  print "<"u+0,length(u+0)">"; 
}

Output

<0 1>
< 0>
<0 1>

Axe

Null pointers can be checked by simply comparing the pointer with 0.

If P=0
 Disp "NULL PTR",i
End

Babel

In this example, we place nil on the stack, then perform an if-then-else (ifte) based on the value returned by the 'nil?' operator which returns true if top-of-stack (TOS) is nil. If TOS is nil, then we can be relieved, otherwise, the interpreter has gone absolutely haywire. The '<<' operator prints the selected string to STDOUT.

{ nil { nil? } { "Whew!\n" } { "Something is terribly wrong!\n" } ifte << }

BASIC

Applesoft BASIC

caveat: http://qconlondon.com/london-2009/presentation/Null+References%3A+The+Billion+Dollar+Mistake

Applesoft has no built-in object system. The closest values to NULL or nil for each of the types are 0 for integers and floating point numbers, and "" for strings. There is also the NUL character: CHR$(0). One could create an object system using global variables and include a special value for NULL, but this is probably a mistake.

TRUE = 1 : FALSE = 0
NULL = TRUE
IF NULL THEN PRINT "NULL"
NULL = FALSE
IF NOT NULL THEN PRINT "NOT NULL"

Output:

NULL
NOT NULL

BBC BASIC

A null object has a pointer with a value of zero or one.

      PROCtestobjects
      END
      
      DEF PROCtestobjects
      PRIVATE a(), b(), s{}, t{}
      DIM a(123)
      DIM s{a%, b#, c$}
      
      IF !^a() <= 1 PRINT "a() is null" ELSE PRINT "a() is not null"
      IF !^b() <= 1 PRINT "b() is null" ELSE PRINT "b() is not null"
      IF !^s{} <= 1 PRINT "s{} is null" ELSE PRINT "s{} is not null"
      IF !^t{} <= 1 PRINT "t{} is null" ELSE PRINT "t{} is not null"
      ENDPROC

Output:

a() is not null
b() is null
s{} is not null
t{} is null

GWBASIC/QBasic/QB/VBDOS

These dialects of BASIC have no built-in object system. One STRING variable can have a default empty ("") value and a numeric one a default zero (0) value. A STRING variable can be assigned with the NULL (Chr$(0)) value if needed and can be assesed with the instruction.

IF VAR$ = CHR$(0) THEN PRINT "Variable has a null value."

SmallBASIC

b = NIL
if(b == NIL) then print "b is not set"

Bracmat

Bracmat has no null objects.

The operators for multiplication, addition and concatenation have neutral elements, which are 1, 0 and the empty string, respectively, but these are values like any other string.

a:?x*a*?z  {assigns 1 to x and to z}
a:?x+a+?z  {assigns 0 to x and to z}
a:?x a ?z  {assigns "" (or (), which is equivalent) to x and to z}

C

C has the null pointer, written as "0", whose internal representation is often, though not always, the same as integer zero. It is (supposedly) garanteed to be pointing to nothing, so receiving one of those likely means you are not looking at an object--but, there are occasions where changing content of a null pointer actually does something (say, on DOS); and a function that's supposed to return a pointer on success doesn't always return a 0 otherwise (e.g. mmap returns -1 for failure).

There is a very common macro, NULL, which evaluates to (void*) 0 or an equivalent value. NULL is compatible with all pointer types, including both data pointers and function pointers.

The standard library defines NULL in locale.h, stddef.h, stdio.h, stdlib.h, string.h, time.h and wchar.h. POSIX systems also define NULL in dirent.h and unistd.h. Many C files include at least one of these headers, so NULL is almost always available.

#include <stdio.h>

int main()
{
	char *object = 0;

	if (object == NULL) {
		puts("object is null");
	}
	return 0;
}

C#

As with Java, any reference type may be null, and testing for nullity uses ordinary boolean operators.

if (foo == null)
    Console.WriteLine("foo is null");

C# 2.0 introduced nullable types for situations in which even primitive value types may have undefined or unknown values (for example, when reading from a database). Prior to the introduction of nullable types, these situations would require writing wrapper classes or casting to a reference type (e.g., object), incurring the penalties of boxing and reduced type safety. A variable with nullable type can be declared simply by adding the '?' operator after the type.

Works with: C# version 2.0+
int? x = 12;
x = null;

Also new in C# 2.0 was the null coalescing operator, '??', which is simply syntactic sugar allowing a default value to replace an operand if the operand is null:

Works with: C# version 2.0+
Console.WriteLine(name ?? "Name not specified");

//Without the null coalescing operator, this would instead be written as:
//if(name == null){
//	Console.WriteLine("Name not specified");
//}else{
//	Console.WriteLine(name);
//}

C++

In C++ non-pointer types do not support null. (C++ provides value semantics rather than reference semantics). When using pointers C++ permits checking for null by comparing the pointer to a literal of 0, or (as in C) by way of a macro (NULL) which simply expands to 0.

#include <iostream>
#include <cstdlib>
if (object == 0) {
   std::cout << "object is null";
}

std::optional is available since C++17 (or Boost's boost::optional via boost/optional.hpp for earlier standards) for cases where the programmer wishes to pass by value, but still support a null value.

#include <iostream>
#include <optional>

std::optional<int> maybeInt()

int main()
{
  std::optional<int> maybe = maybeInt();

  if(!maybe)
    std::cout << "object is null\n";
}

C++11

In C++11 there is nullptr of type nullptr_t which represents a pointer to an invalid place. You can use it like

int *p = nullptr;
...
if (p == nullptr){
  // do some thing
}
//or just
if (p){
  // do some thing
}

Chapel

Class variables can be declared as nil if and only if they have nilable class type, declared with postfix ?, alongside memory management technique owned, borrowed, shared, or unmanaged. A nilable class type is default initialized to nil.

class C { };
var c : owned C?; // is nil
writeln(if c == nil then "nil" else "something");

Clojure

Clojure's nil is equivalent to Java's null.

(let [x nil]
 (println "Object is" (if (nil? x) "nil" "not nil")))

Test wether symbol foo is defined:

(find (ns-interns *ns*) 'foo)

Undefining foo:

(ns-unmap *ns* 'foo)

COBOL

Works with GnuCOBOL 2.0

       identification division.
       program-id. null-objects.
       remarks. test with cobc -x -j null-objects.cob

       data division.
       working-storage section.
       01 thing-not-thing      usage pointer.

      *> call a subprogram
      *>   with one null pointer
      *>   an omitted parameter
      *>   and expect void return (callee returning omitted)
      *>   and do not touch default return-code (returning nothing)
       procedure division.
       call "test-null" using thing-not-thing omitted returning nothing
       goback.
       end program null-objects.

      *> Test for pointer to null (still a real thing that takes space)
      *>   and an omitted parameter, (call frame has placeholder)
      *>   and finally, return void, (omitted)
       identification division.
       program-id. test-null.

       data division.
       linkage section.
       01 thing-one            usage pointer.
       01 thing-two            pic x.

       procedure division using
           thing-one
           optional thing-two
           returning omitted.

       if thing-one equal null then
           display "thing-one pointer to null" upon syserr
       end-if

       if thing-two omitted then
           display "no thing-two was passed" upon syserr
       end-if
       goback.
       end program test-null.
Output:
prompt$ cobc -x -j null-objects.cob
thing-one pointer to null
no thing-two was passed

Common Lisp

Basics

Common Lisp has an object denoted by the symbol nil. When the symbol nil is evaluated as an expression, it evaluates to itself.

nil uniquely represents boolean false, and so code like

(if (condition) (do-this))

is actually testing whether (condition) returns the value nil. The object nil is also used to denote the empty list which also terminates other lists. The value is also used as a default when some function returns fewer values than expected. (list (values)) produces (nil) (list containing one element, which is the empty list), because (values) produces no value, but the function call (list ...) needs to reduce the expression to a single argument value, and so nil is supplied.

Beginnings of Null Object

The idea of making functions accept nil without failing did not appear in early Lisps. For instance (car nil) was erroneous: it was incorrect to try to access the first element of a non-list.

The defaulting behavior (car nil) which Common Lisp programmers take for granted was introduced in InterLisp, and then copied into MacLisp. (InterLisp had other liberties that do not survive into Common Lisp: it was possible to call a function with insufficient arguments, and the missing ones defaulted to nil. Likewise, excess arguments were ignored. CL has a disciplined syntax and semantics for default and variable arguments.)

This (car nil) -> nil behavior shows nil in an kind of new role: the role of a null object which takes methods that apply to other objects and provides some default non-failing behavior. It is the beginnings of the [null object design pattern].

Object-Oriented Null Object

In Common Lisp, in fact, there is a class called null, of which the object nil is understood to be the only instance. Furthermore, the null class is at the bottom of the type spindle: it is a subclass of every class. This is in contrast with the type T which is a superclass of every class.

Since null is at the bottom of the class hierarchy, it is possible to write methods specialized to parameters of class null which will only be applicable if the argument is the object nil. No other object is a subtype of null.

Some traditional Lisp functions could be expressed using the object system like this. Suppose that the car function did not have a safe defaulting behavior for nil. We could use the methods of the object system to define a car* which does have the safe behavior:

(defmethod car* ((arg cons))
  (car arg))

(defmethod car* ((arg null))
  nil)

Now if we invoke car* on something which is neither a cons, nor nil, we get an error about no applicable method being found.

We can handle that ourselves by writing a method specialized to the master supertype t:

(defmethod car* ((arg t))  ;; can just be written (defmethod car* (arg) ...)
  (error "CAR*: ~s is neither a cons nor nil" arg))

The classes t and null are widely exploited in Lisp OO programming.

Component Pascal

MODULE ObjectNil;
IMPORT StdLog;
TYPE 
	Object = POINTER TO ObjectDesc;
	ObjectDesc = RECORD
	END;
VAR	
	x: Object; (* default initialization to NIL *)
	
PROCEDURE DoIt*;
BEGIN 
	IF x = NIL THEN
		StdLog.String("x is NIL");StdLog.Ln
	END
END DoIt;

END ObjectNil.

Crystal

In Crystal, nil is represented by an instance of the Nil type, accessed by the identifier nil. A variable can only become nil if Nil is one of its possible types. All objects inheriting from the base Object class implement the method .nil? which returns true if the object is nil and false if it isn't. The equality and case equality operators can also be used to check for nil. The compiler returns an error if an object may be nil but is not treated as such. This can be suppressed with the .not_nil! method, which throws an exception at runtime if the object is in fact nil.

foo : Int32 | Nil = 5 # this variable's type can be Int32 or Nil
bar : Int32? = nil    # equivalent type to above, but shorter syntax
baz : Int32 = 5       # this variable can never be nil

foo.not_nil!          # nothing happens, since 5 is not nil
puts "Is foo nil? #{foo.nil?}"
foo = nil
puts "Now is foo nil? #{foo.nil?}"

puts "Does bar equal nil? #{bar == nil}"

puts "Is bar equivalent to nil? #{bar === nil}"

bar.not_nil!          # bar is nil, so an exception is thrown
Output:
Is foo nil? false
Now is foo nil? true
Does bar equal nil? true
Is bar equivalent to nil? true
Unhandled exception: Nil assertion failed (NilAssertionError)
...


D

In D is is used to perform bitwise identity, like to compare an object reference against null.

import std.stdio;

class K {}

void main() {
    K k;
    if (k is null)
        writeln("k is null");
    k = new K;
    if (k !is null)
        writeln("Now k is not null");
}
Output:
k is null
Now k is not null

Delphi

  // the following are equivalent
  if lObject = nil then
  ...
  
  if not Assigned(lObject) then 
  ...

DuckDB

Works with: DuckDB version V1.0

DuckDB follows SQL in matters of `NULL` but it is important to note that DuckDB also follows JSON regarding the JSON value, `null`. This can sometimes be a bit confusing, because the JSON `null` value generally prints in the same way as the VARCHAR value `'null'`. Hopefully the following examples will clarify whatever is not already clear. The "D " signifies the DuckDB prompt.

JSON `null`

D select NULL, 'null', 'null'::JSON;
┌───────┬─────────┬────────────────────────┐
│ NULL  │ 'null'  │ CAST('null' AS "JSON") │
│ int32 │ varchar │          json          │
├───────┼─────────┼────────────────────────┤
│       │ null    │ null                   │
└───────┴─────────┴────────────────────────┘

Note that, despite the left-hand column in the table above, the DuckDB type of `NULL` is actually the string '"NULL"':

D select typeof(NULL), typeof(NULL) = '"NULL"', typeof('null'), typeof('null'::JSON);
┌──────────────┬───────────────────────────┬────────────────┬────────────────────────────────┐
│ typeof(NULL) │ (typeof(NULL) = '"NULL"') │ typeof('null') │ typeof(CAST('null' AS "JSON")) │
│   varchar    │          boolean          │    varchar     │            varchar             │
├──────────────┼───────────────────────────┼────────────────┼────────────────────────────────┤
│ "NULL"       │ true                      │ VARCHAR        │ JSON                           │
└──────────────┴───────────────────────────┴────────────────┴────────────────────────────────┘

NULL

D select NULL, NULL=NULL, NULL IS NULL, NULL IS NOT NULL, COALESCE(NULL,0);

┌───────┬───────────────┬────────────────┬────────────────────┬───────────────────┐
│ NULL  │ (NULL = NULL) │ (NULL IS NULL) │ (NULL IS NOT NULL) │ COALESCE(NULL, 0) │
│ int32 │    boolean    │    boolean     │      boolean       │       int32       │
├───────┼───────────────┼────────────────┼────────────────────┼───────────────────┤
│       │               │ true           │ false              │                 0 │
└───────┴───────────────┴────────────────┴────────────────────┴───────────────────┘

Note however that neither [NULL] nor (NULL,NULL) is NULL:

D select [NULL] IS NULL,  [NULL][1] IS NULL;
┌─────────────────────────────────┬────────────────────────────────────┐
│ (main.list_value(NULL) IS NULL) │ (main.list_value(NULL)[1] IS NULL) │
│             boolean             │              boolean               │
├─────────────────────────────────┼────────────────────────────────────┤
│ false                           │ true                               │
└─────────────────────────────────┴────────────────────────────────────┘

D select (NULL) IS NULL, (NULL,NULL), (NULL,NULL) IS NULL, (NULL,NULL)[1] IS NULL;
┌────────────────┬──────────────────────────┬──────────────────────────────────┬─────────────────────────────────────┐
│ (NULL IS NULL) │  main."row"(NULL, NULL)  │ (main."row"(NULL, NULL) IS NULL) │ (main."row"(NULL, NULL)[1] IS NULL) │
│    boolean     │ struct(integer, integer) │             boolean              │               boolean               │
├────────────────┼──────────────────────────┼──────────────────────────────────┼─────────────────────────────────────┤
│ true           │ (NULL, NULL)             │ false                            │ true                                │
└────────────────┴──────────────────────────┴──────────────────────────────────┴─────────────────────────────────────┘

DWScript

See Delphi

Dyalect

Dyalect has a notion of nil - a special sigleton value which can be used in the cases when no other meaningful value can be provided.

var x = nil
if x == nil {
  //Do something
}

Déjà Vu

There isn't an actual null object, so generally falsy objects are used to indicate a missing value, or when that's impractical a specific ident:

if not obj:
    pass #obj is seen as null

if = :nil obj:
    pass #obj is seen as null

E

object == null

EchoLisp

The null object - null - is the same as the empty list (). It may be tested with the null? or !null? predicates. NB : null is not the same as the boolean #f (false). null evaluates to #t (true) in logical operations.

null  null
()  null
(null? 3)  #f
(!null? 4)  #t
(null? null)  #t

;; careful - null is not false :
(if null 'OUI 'NON)  OUI

;; usual usage : recursion on lists until (null? list)
(define (f list) 
    (when (!null? list) 
    (write (first list)) (f (rest list))))

(f '( a b c))    a b c

Ecstasy

In Ecstasy, everything is an object, including the  Null  value.  Null  is the only value in the  Nullable  enumeration defined in the "ecstasy" core module. As a regular old object, the  Null  value has a regular old class, a regular old type, and it implements regular old methods such as  toString() . There are, however, a few specific ways in the language (both the compiler and runtime) that  Null  is treated specially:

  • Syntax support: A type union with  Nullable  can be indicated with the postfix "?"; for example, the long-hand type union syntax  Nullable|String s  can be replaced using the short-hand notation  String? s .
  • Syntax support: The postfix "?" operator is a short-circuiting null value test, allowing a cascading sequence of null tests to replace a series of nested if statements; for example,  Int x = a?.b()?.c?[i?].d? : e;  will result in the value of  e  if any of the expressions  a ,  a.b() ,  a.b().c ,  i , or  a.b().c[i].d  are  Null .
  • Syntax support: The expression  x ?: y  will use the value  y  iff  x  is  Null , and the corresponding assignment statement  x ?:= y  will assign  y  to  x  iff  x  is  Null .
  • Syntax support: The assignment statement  x ?= y  will assign  y  to  x  iff  y  is not  Null .
  • The assignment operator  ?=  also yields a  Boolean  value indicating the non-null-ness the right hand side value, which can be used in an  if  condition,  while  condition, etc.; for example,  if (x ?= y) ...  will take the "then" branch if  y  is non-null and therefore  x  is definitely assigned, otherwise it will take the "else" branch and  x  will not be definitely assigned.
  •  Null  values are treated specially by equality comparisons: Normally, the compiler prevents two references of different compile-time types from being compared (such as  Int  and  String ), but an explicit exception is made that allows a nullable type to be compared with a non-nullable type (such as  String?  and  String ).

Other than these specific compiler and runtime features,  Null  is treated exactly like every other object.


module NullObject {
    void run() {
        @Inject Console console;
        console.print($"Null value={Null}, Null.toString()={Null.toString()}");

     // String s = Null;        // <-- compiler error: cannot assign Null to a String type
        String? s = Null;       // "String?" is shorthand for the union "Nullable|String"
        String s2 = "test";
        console.print($"{s=}, {s2=}, {s==s2=}");

     // Int len = s.size;       // <-- compiler error: String? does not have a "size" property
        Int len = s?.size : 0;
        console.print($"{len=}");

        if (String test ?= s) {
            // "s" is still Null in this test, we never get here
        } else {
            s = "a non-null value";
        }

     // if (String test ?= s){} // <-- compiler error: The expression type is not nullable
        s2 = s;                 // at this point, s is known to be a non-null String
        console.print($"{s=}, {s2=}, {s==s2=}");
    }
}
Output:
Null value=Null, Null.toString()=Null
s=Null, s2=test, s==s2=False
len=0
s=a non-null value, s2=a non-null value, s==s2=True

Eiffel

Any reference type variable can be Void. In the following example, STRING is a reference type, while INTEGER is an expanded type. The keyword "detachable" (as opposed to "attached") is used to indicate that the variable "s" may be Void. The default interpretation when neither of these two keywords is used depends on a compiler option. The first if statement will cause a compiler warning because an expanded type variable such as i will never be Void.

class
	APPLICATION
inherit
	ARGUMENTS
create
	make

feature {NONE} -- Initialization

	make
		local
			i: INTEGER
			s: detachable STRING
		do
			if i = Void then
				print("i = Void")
			end
			if s = Void then
				print("s = Void")
			end
		end
end
Output:
s = Void

Elixir

nil is atom in fact:

iex(1)> nil == :nil
true
iex(2)> is_nil(nil)
true

nil is thought of as being false in the conditional expression.

If the condition given to if/2 returns false or nil, the body given between do/end is not executed and it simply returns nil.

iex(3)> if nil, do: "not execute"
nil

EMal

^|
 | EMal has the Variable type (and its keyword var) that is the nullable universal supertype.
 | EMal has the Void type (and its keyword void) that holds only one value: null.
 | EMal has not nullable types (logic, int, real, text, blob), but null equality is always allowed.
 |^
var a # defaults to null
int b # defaults to 0
void c # only one allowed value: null
writeLine("nullable var equals to not nullable int: " + (a == b)) # allowed, false
^| if the data type of a is void we are sure that a is null |^
writeLine("type of a equals to Void data type: " + (generic!a == void)) # true
writeLine("integer value " + b + " equals to null: " + (b == null)) # allowed, always false
writeLine("a void value equals to null: " + (c == null)) # always true
Output:
nullable var equals to not nullable int: ⊥
type of a equals to Void data type: ⊤
integer value 0 equals to null: ⊥
a void value equals to null: ⊤

Erlang

Erlang does not have an null object.

As an alternative, many applications tend to pick a convention for returning an empty condition and use that.

Example alternatives:

  1. Something like
    {ok, 3} % normal case
    or
    {err, no_more} % error case
    on error.
  2. Don't ever allow an undefined return value, and throw an exception instead.
  3. Return an atom:
    1. undefined*
    2. undef
    3. null
    4. nil
    5. none

undefined is often used by records as an initial value and the stdlib module.

Atoms are erlang's user-defined constants that always evaluates to is itself. It is also equal to no other value else but itself.

F#

As a .Net languages F# inherits the null as a potential value for object variables. Other than in interfacing assemblies written in other .Net languages, null rarely serves a purpose in F# code. Contrived code, to show using null, as per task description:

let sl : string list = [null; "abc"]

let f s =
    match s with
    | null -> "It is null!"
    | _ -> "It's non-null: " + s

for s in sl do printfn "%s" (f s)

Factor

: is-f? ( obj -- ? ) f = ;

Fantom

Test for equality with 'null', which is the null value.

fansh> x := null
fansh> x == null
true
fansh> x = 1
1
fansh> x == null
false

Note, nullable objects have a type ending in a question mark, for example:

Int? y := null is valid, but

Int y := null is not.

Forth

Standard ANS Forth does not distinguish a particular invalid memory value like NULL. Instead, ALLOCATE returns an out-of-band success code to indicate a failed allocation. Dictionary words have the option of throwing an exception on a dictionary space overrun. Forth lacks a NULL symbol because it has such a wide variety of target platforms. On some embedded targets, the memory space may be as small as 64 direct-mapped addresses, where eliminating a valid zero address would have a high price.

In practice, all notable hosted implementations follow the C practice of being able to treat a zero address (i.e. FALSE) as a null address for the purpose of list termination.

FreeBASIC

'FB 1.05.0 Win64

' FreeBASIC does not have a NULL keyword but it's possible to create one using a macro

#Define NULL CPtr(Any Ptr, 0) '' Any Ptr is implicitly convertible to pointers of other types

Type Dog
  name As String
  age As Integer
End Type

Dim d As Dog Ptr = New Dog
d->Name = "Rover"
d->Age = 5
Print d->Name, d->Age
Delete d 
d = NULL '' guard against 'd' being used accidentally in future 

' in practice many FB developers would simply have written: d = 0 above
Sleep
Output:
Rover          5


FutureBasic

While objects such as strings can be NULL in FB, arrays, dictionaries and other collections cannot contain NULL objects.

// Object dimensioned, but not assigned
CFStringRef object

if ( object == NULL )
print "object is NULL"
end if

HandleEvents
Output:
object is NULL

GDScript

Godot has a null value. Here is an example of dealing with null.

extends Node2D

func _ready() -> void:
	var empty : Object
	var not_empty = Object.new()
	
	# Compare with null. 
	if empty == null:
		print("empty is null")
	else:
		print("empty is not null")
	
	# C-like comparation.
	if not_empty:
		print("not_empty is not null")
	else:
		print("not_empty is null")
	return
Output:
empty is null
not_empty is not null

Go

Nil is a predefined identifier, defined for six types in Go. In each case, it represents the zero value for the type, that is, the memory representation of all zero bytes. This is the value of a newly created object. In the cases of these six types, an object must be subsequently initialized in some way before it has much use. Examples of initialization are given in the Go solution of task Undefined values.

package main

import "fmt"

var (
    s []int       // slice type
    p *int        // pointer type
    f func()      // function type
    i interface{} // interface type
    m map[int]int // map type
    c chan int    // channel type
)

func main() {
    fmt.Println(s == nil)
    fmt.Println(p == nil)
    fmt.Println(f == nil)
    fmt.Println(i == nil)
    fmt.Println(m == nil)
    fmt.Println(c == nil)
}

Output is "true" in each case.

Haskell

Haskell does not have a universal null value. There is a 'value of every type', the undefined value (sometimes written ⊥, 'bottom'), but it is essentially a sort of exception — any attempt to use it is an error.

undefined      -- undefined value provided by the standard library
error "oops"   -- another undefined value
head []        -- undefined, you can't take the head of an empty list

When one would use "null" as a marker for "there is no normal value here" (e.g. a field which is either an integer or null), one uses the Maybe type instead. The definition of Maybe is:

 data Maybe a = Nothing | Just a

That is, a Maybe Integer is either Nothing or Just <some integer>.

There are many ways to work with Maybe, but here's a basic case expression:

case thing of
 Nothing -> "It's Nothing. Or null, whatever."
 Just v  -> "It's not Nothing; it is " ++ show v ++ "."

It is easy to work with Maybe type using do-notation (since Maybe is a monad):

add_two_maybe_numbers x y do
  a <- x
  b <- y
  return (a+b)

Then

*Main> add_two_maybe_numbers (Just 2) (Just 3)
Just 5
*Main> add_two_maybe_numbers (Just 2) Nothing 
Nothing

Icon and Unicon

Icon/Unicon have a null value/datatype. It isn't possible to undefine a variable.

procedure main()
nulltest("a",a)                # unassigned variables are null by default
nulltest("b",b := &null)       # explicit assignment is possible
nulltest("c",c := "anything")
nulltest("c",c := &null)       # varibables can't be undefined
end

procedure nulltest(name,var)
return write(name, if /var then " is" else " is not"," null.")
end

Io

if(object == nil, "object is nil" println)

J

J doesn't have an untyped NULL. Instead, it has a concept of "fill". Numeric fill is 0, character fill is the space character, and boxed fill is the ace (a:) which is an empty box. Fill is what is used to pad an array structure when that is needed. (And some operations support using a user specified value in place of the default fill.)

To indicate "missing data", "normal" data is usually pressed into service (e.g. 0 or _1 in a numeric context, ' ' in a literal context, a: in a boxed context, etc). Frequently, missing data is represented by the empty vector '', or other arrays without any elements.

That said, undefined names in J are not associated with any data of any type. Furthermore, any attempt to use the value of an undefined is treated as an error (this is distinct from the concept of an empty array, which contains no data but which is not an error to use). However, it is possible to check if a name is defined before attempting to use it:

isUndefined=: _1 = nc@boxxopen

Example use:

   isUndefined 'foo'
1
   foo=:9
   isUndefined 'foo'
0

Note, of course, that this "name is not defined" state is not a first class value in J -- you can not create a list of "undefineds".

Finally, note: the concept of an empty array can be natural in J (and APL) for representing data which is not there -- it is the structural equivalent of the number zero. That said, its implications can sometimes be non-obvious for people coming from a languages which requires that arrays have content. As a result, you will sometimes encounter empty array jokes...

Marie Pennysworth, having spent a productive day shopping, stopped by Robert Cuttingham's butcher shop.
"How much for your t-bones?" she asked.
"Eleven dollars per pound," he responded.
"How about for your sirloin?" she continued.
"Sirloin is thirteen dollars per pound today," he answered.
"But Harkin's Grocery down the street is selling sirloin for nine dollars per pound!" she exclaimed.
"So, why don't you buy it from them?" he asked.
"Well, they're out," she sighed.
He smiled, "When I am out, I only charge seven dollars a pound."

That said, note that a typical way to indicate missing or invalid data, in J, is to have a parallel array which is a bit mask (which selects the desired or valid values and, by implication, does not select the invalid values). Or, as a logical equivalent: a list of indices which select the desired and/or valid values. Alternatively, you can have an array without the invalid values and a bit mask which demonstrates how the data would be populated on a larger array -- in other words instead of 3,4,null,5 you could have (3 4 5) and (1 1 0 1). And you can transform between some of these representations:

   1 1 0 1#3 4 _ 5           NB. use bitmask to select numbers
3 4 5
   I.1 1 0 1                 NB. get indices for bitmask
0 1 3
   0 1 3 { 3 4 _ 5           NB. use indices to select numbers
3 4 5
   1 1 0 1 #inv 3 4 5        NB. use bitmask to restore original positions
3 4 0 5
   1 1 0 1 #!._ inv 3 4 5    NB. specify different fill element
3 4 _ 5
   3 4 5 (0 1 3}) _ _ _ _    NB. use indices to restore original positions
3 4 _ 5

Java

In Java, "null" is a value of every reference type.

// here "object" is a reference
if (object == null) {
   System.out.println("object is null");
}

JavaScript

In Javascript null is the value that isn't anything. null is not an object, but because of a bug typeof null will return "object".

if (object === null) {
  alert("object is null");
  // The object is nothing
}

typeof null === "object"; // This stands since the beginning of JavaScript

jq

jq has null as a value, but while on the subject of nothing, it may be worth mentioning that jq also has a filter, empty, for producing an empty sequence, a.k.a. nothing.

null is distinct from false. Here are some examples:

null|type      # => "null"

null == false  # => false

null == null   # => true

empty|type     # =>  # i.e. nothing (as in, nada)

empty == empty # =>  # niente

empty == "black hole" # =>  # Ничего

Jsish

Like Javascript, Jsish has undefined and null. Unlike Javascript, null is not typed as object, but null.

Jsish, with parameter typed functions, also allows void as a type spec, to indicate the parameter (of whatever type) may be omitted by a caller.

/* null non value */

if (thing == null) { puts("thing tests as null"); }
if (thing === undefined) { puts("thing strictly tests as undefined"); }
puts(typeof thing);
puts(typeof null);
puts(typeof undefined);
Output:
prompt$ jsish nulling.jsi
thing tests as null
thing strictly tests as undefined
undefined
null
undefined

Julia

See language reference: https://docs.julialang.org/en/v1/manual/faq/#Nothingness-and-missing-values

K

K has well developed notions of data null :

The special numeric atoms 0I and 0N refer to integer infinity and “not-a-number” (or “null” in database parlance) concepts, and similarly 0i and 0n for floating-point.

eg : ( 1 2 3 0N 6 7 )

and and missing value nil :

Empty expressions in both list expressions and function expressions actually represent a special atomic value called nil. ... A list may contain one or more empty items (i.e. the nil value _n), which are typically indicated by omission:
  (1;;2) ~ (1 ; _n ; 2)    /  ~  is ''identical to'' or ''match'' .
1
  _n ~' ( 1 ; ; 2 )        /  ''match each''
0 1 0 

additional properties :  _n@i and _n?i are i; _n`v is _n

For more detail on K's concept of typed nulls, see http://code.kx.com/wiki/Reference/Datatypes#Primitive_Types

Klingphix

%t nan !t
$t nan == ?
Output:
1

Komodo

Komodo has a unit type, represented by an empty tuple. You can check for it with patterns or the equality operator:

let unit := ()
let isUnit(()) := true
let isUnit(_) := false

assert(isUnit(unit))
assert(unit = ())

Kotlin

Kotlin distinguishes between non-nullable types and nullable types. The latter are distinguished from the former by a '?' suffix. Only nullable types have a 'null' value indicating that they don't currently refer to an object of their non-nullable equivalent.

In addition, Kotlin has a Nothing type which has no instances and is a sub-type of every other type. There is also a nullable Nothing? type whose only value is 'null' and so, technically, this is the type of 'null' itself.

Here are some examples:

// version 1.1.0

fun main(args: Array<String>) {
    val i: Int  = 3           // non-nullable Int type - can't be assigned null
    println(i)
    val j: Int? = null        // nullable Int type - can be assigned null
    println(j) 
    println(null is Nothing?) // test that null is indeed of type Nothing? 
}
Output:
3
null
true

langur

Null can be compared for directly, using equality operators. Operators ending with a ? mark propagate null. A null in an expression test is a non-truthy result.

var x, y = true, null

writeln x == null
writeln y == null
writeln x ==? null
writeln y ==? null

# null not a "truthy" result
writeln if(null: 0; 1)
Output:
false
true
null
null
1

Lasso

local(x = string, y = null)
#x->isA(::null) 
// 0 (false)

#y->isA(::null) 
// 1 (true)

#x == null
// false

#y == null
//true

#x->type == 'null'
// false

#y->type == 'null'
//true

Latitude

Nil is an object in Latitude, like any other.

foo := Nil.
if { foo nil?. } then {
  putln: "Foo is nil".
} else {
  putln: "Foo is not nil".
}.

In particular, Nil satisfies the Collection mixin, so it can be treated as an (immutable) collection.

Nil to (Array). ;; []

Nil is the default value returned if a method body is empty.

func := {}.
func. ;; Nil

Lily

Lily doesn't provide a built-in nothing type, but allows one to be created using enum class:

enum class Option[A] {
  Some(A)
  None
}

# Only variables of class Option can be assigned to None.

# Type: Option[integer]
var v = Some(10)

# Valid: v is an Option, and any Option can be assigned to None
v = None

# Invalid! v is an Option[integer], not just a plain integer.
v = 10

# Type: integer
var w = 10

# Invalid! Likewise, w is an integer, not an Option.
w = None

Lingo

Null/nil is called "<Void>" in Lingo. Lingo doesn't distinguish undefined variables from <Void> objects, and by using the constant VOID you can even assign <Void> to variables. Functions that don't return anything, return <Void>. Checking for <Void> (e.g. by using built-in function voidP) can be used to implement optional function arguments: if voidP() returns TRUE (1) for some argument, a default value can be assigned in the function body.

put _global.doesNotExist
-- <Void>

put voidP(_global.doesNotExist)
-- 1

x = VOID
put x
-- <Void>

put voidP(x)
-- 1

to test :thing
if empty? :thing [print [list or word is empty]]
end

print empty? []  ; true
print empty? "|| ; true

Lua

isnil = (object == nil)
print(isnil)

M2000 Interpreter

For Com Objects

There is a Nothing to assign to a COM object to released (but time to actually released depends from system). A com pointer can't get another value (only the first value, and the Nothing at the end).

Module CheckWord {
      Declare Alfa "WORD.APPLICATION"
      Declare Alfa Nothing
      Print Type$(Alfa)="Nothing"
      Try ok {
            Declare Alfa "WORD.APPLICATION"
            \\ we can't declare again Alfa      
       
      }
      If Not Ok Then Print Error$  ' return Bad Object declaration
}
CheckWord

For Containers

Container's pointers (for arrays, inventories, stack) we have to assign an empty container, there is not a null one.

Module CheckContainers {
      \\ Arrays  (A() and B() are value types)
      Dim A(10)=1, B()
      \\ B() get a copy of A(), is not a reference type
      B()=A()
      \\ we make a pointer to Array
      B=A()
      \\ now B is a reference type object
      Print Len(B)=10  ' ten items
      B+=10
      Print A(3)=11, A(7)=11
      \\ we can change pointer using a pointer to an empty array
      B=(,)
      \\ we can erase A() and B()
      Dim A(0), B(0)
      Print Len(A())=0, Len(B())=0
      Print Len(B)=0 
      B=(123,)
      \\ B() is a value type so get a copy
      B()=B
      Print Len(B)=1, B(0)=123
      \\ Using Clear we pass a new empty array
      Clear B
      Print Type$(B)="mArray"
      Print Len(B)=1, B(0)=123
      
      
      \\ Inventrories. Keys must be unique (for normal inventories)
      Inventory M=1,2,3,4:=400,5
      Print M
      Clear M
      Inventory M=1,2,3,4,5
      Print M
      
      \\ Inventory Queue can have same keys.
      Inventory Queue N=1,1,2:="old",2:="ok",3
      If Exist(N,2) Then Print Eval$(N)="ok", Eval(N!)=3 ' 4th item
      Clear N
      Print Len(N)=0, Type$(N)="Inventory"
      
      \\ Stack Object
       Z=Stack:=1,2,3
       Stack Z {
            While not empty {Print Number}
      }
      Print Len(Z)=0
      Z=Stack((Stack:=1,2,3,4),Stack:=20,30,40 )
      Print Len(Z)=7
      Print Z    ' 1 2 3 4 20 30 49
      Z=Stack  ' This is an empty stacl
      Print Len(Z)=0
      Print Type$(Z)="mStiva"
}
CheckContainers

For Groups

Groups are value types, but we can make reference to them,or pointer to them

A Named referenced can't get a new reference

A pointer to a named group is actual a reference, and can change type and reference

A pointer to a copy of group (as float group) is actually a pointer to group.

Pointers to groups can be point to an Null Group, assigning a 0& value (a long type)


class something {
}
class alfa as something {
	x=10, y=20
}
a->alfa()
Print a is type alfa = true
Print a is type something = true
a->0&
Print a is type null = true
\\ beta is a named object, is static
group beta {
type: something, alfa
	x=10, y=20
}
Print beta is type alfa = true
Print beta is type something = true
\\ now a is a pointer as a weak reference to beta
a->beta
print a is type alfa = true
print a is type something = true
a=pointer()   ' same as a->0&
Print a is type null = true
\\ now a is a pointer of a copy of beta
a->(beta)
print a is type alfa = true
print a is type something = true
a=pointer()   ' same as a->0&
Print a is type null = true

Maple

In Maple, NULL and () represent the null object.

a := NULL;
                                                         a := 
is (NULL = ());
                                                         true
if a = NULL then
    print (NULL);
end if;

A null object is different from an undefined value.

b := Array([1, 2, 3, Integer(undefined), 5]);
                                                        b := [ 1 2 3 undefined 5 ]
numelems(b);
                                                        5
b := Array([1, 2, 3, Float(undefined), 5]);
                                                        b := [ 1 2 3 Float(undefined) 5 ]
numelems(b);
                                                        5
b := Array([1, 2, 3, NULL, 5]);
                                                        b := [ 1 2 3 5 ]
numelems(b);
                                                        4

Mathematica /Wolfram Language

Mathematica can assign a Null value to a symbol, two examples:

x=Null;
x =.
x = (1 + 2;)
FullForm[x]

Both set x to be Null. To specifically test is something is Null one can use the SameQ function (with infix operator: ===):

SameQ[x,Null]

Or equivalent:

x===Null

will give back True if and only if x is assigned to be Null. If x is empty (nothing assigned) this will return False. To test if an object has something assigned (number, list, graphics, null, infinity, symbol, equation, pattern, whatever) one uses ValueQ:

x =.;
ValueQ[x]
x = 3;
ValueQ[x]

gives:

False
True

MATLAB / Octave

The closest think to a NULL element in Matlab/Octave is an empty field or empty string; empty fields in a conditional expression evaluate to false.

a = []; b='';
isempty(a)
isempty(b)
if (a) 
  1, 
else, 
  0
end;
octave:4> a = []; b='';
octave:5> isempty(a)
ans =  1
octave:6> isempty(b)
ans =  1
octave:7> if (a) 1, else, 0, end;
ans = 0

Maxima

There is no null object in Maxima. Usually, a function that returns nothing (as the builtin "disp") returns in fact the symbol 'done.

MAXScript

if obj == undefined then print "Obj is undefined"

min

null null? puts!
Output:
true

Modula-3

In Modula-3, NIL is a value, and NULL is a type. The NULL type contains only one value, NIL. NULL is a subtype of all reference types, which allows all reference types to have the value NIL.

This can lead to errors, if for example you write:

VAR foo := NIL

This (most likely incorrectly) gives foo the type NULL, which can only have the value NIL, so trying to assign it anything else will not work. To overcome this problem, you must specify the reference type when declaring foo:

VAR foo: REF INTEGER := NIL;
IF foo = NIL THEN
  IO.Put("Object is nil.\n");
END;

MUMPS

A variable can be declared implicitly by using it as on the left side in a SET, or by making a new version for the current scope with a NEW statement. A variable can have descendants without having a value set.

The $DATA (or $D) function will return a number:

$DATA returns: Variable is defined
Variable has children   No Yes
No 0 1
Yes 10 11

Or, by examples (in immediate mode):

CACHE>WRITE $DATA(VARI)
0
CACHE>SET VARI="HELLO" WRITE $DATA(VARI)
1
CACHE>NEW VARI WRITE $DATA(VARI) ;Change to a new scope
0
CACHE 1S1>SET VARI(1,2)="DOWN" WRITE $DATA(VARI)
10
CACHE 1S1>WRITE $DATA(VARI(1))
10
CACHE 1S1>WRITE $D(VARI(1,2))
1
CACHE 1S1>SET VARI(1)="UP" WRITE $DATA(VARI(1))
11
<CACHE 1S1>QUIT ;Leave the scope

<CACHE>W $DATA(VARI)," ",VARI
1 HELLO

Nanoquery

Translation of: Ursa
$x = $null

if ($x = $null)
	println "x is null"
else
	println "x is not null"
end if

Neko

/**
 <doc>
 <p>Neko uses <i>null</i> for undefined variables,
  and also as a programmer accessible value.</p>
 <p>The <i>null</i> value can be treated as a boolean value with the
  builtin $istrue, and tests as false.</p>
 </doc>
*/

var n = null
if n == null $print("n is null\n")
if $not($istrue(n)) $print("and tests as boolean false\n")

NetRexx

In NetRexx as in Java, "null" is a value of every reference type.

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

robject = Rexx -- create an object for which the value is undefined
say String.valueOf(robject) -- will report the text "null"
if robject = null then say 'Really, it''s "null"!'

Output:

null 
Really, it's "null"!

NewLISP

#! /usr/local/bin/newlisp
(setq myobject nil)
(println (nil? myobject))
(exit)
true

Nim

There is a nil value in Nim, which is the same as a 0. It can be explicitly forbidden as a value:

let s: pointer = nil

{.experimental: "notnil".}
let ns: pointer not nil = nil # Compile time error

The value "nil" can be used for pointers, references (i.e. pointers managed by the garbage collector) and procedures. It was also used for strings and sequences, but this is no longer the case (option --nilseqs:on allows to retrieve the old behavior).

Testing if a pointer “p” is nil can be done either by using == or using the procedure isNil.

var p: ptr int
if p == nil: echo "it is nil"
if p != nil: echo "it is not nil"
if p.isNil: echo "it is nil"

Nu

The command ignore and the literal null both return 'nothing'.

let $var = ignore
$var == null
Output:
true

Oberon-2

Works with: oo2c
MODULE Null;
IMPORT
  Out;
TYPE
  Object = POINTER TO ObjectDesc;
  ObjectDesc = RECORD
  END;

VAR
  o: Object;  (* default initialization to NIL *)

BEGIN
  IF o = NIL THEN Out.String("o is NIL"); Out.Ln END
END Null.
Output:
o is NIL

Objeck

In Objeck, "Nil" is a value of every reference type.

# here "object" is a reference
if(object = Nil) {
   "object is null"->PrintLine();
};

Objective-C

The value nil is used to indicate that an object pointer (variable of type id) doesn't point to a valid object.

// here "object" is an object pointer
if (object == nil) {
   NSLog("object is nil");
}

An interesting thing is that in Objective-C, it is possible to send a message to nil, and the program will not crash or raise an exception (nothing will be executed and nil will be returned in place of the usual return value).

[nil fooBar];

Note that nil is distinct from NULL, which is only used for regular C pointers.

For class pointers (values of type Class), they have a separate null pointer value called Nil.

Confusingly, there is also NSNull, a singleton class with one value, [NSNull null], used as a dummy object to represent the lack of a useful object. This is needed in collections like arrays and dictionaries, etc., because they do not allow nil elements, so if you want to represent some "empty" slots in the array you would use this.

OCaml

Maybe the closest type of OCaml would be the type option, which is defined like this in the standard library:

type 'a option = None | Some of 'a
match v with
| None -> "unbound value"
| Some _ -> "bounded value"

Oforth

null is an object, the only instance of Null class. When an object is created, all attributes are initiallized to null value. When a method or function is called, all local variables begin with null value.

null isNull
"abcd" isNull
: testNull { | a | a ifNull: [ "Variable value is null" println ] ;

Ol

Ol has no null object in task meaning sense. To indicate the "unassigned" variable state typically used #false because this is only value that triggers the 'unless' and 'if not'.

The builtin null and #null (that a same) means "an empty list". Important note: in contrast with CL the null means #true in 'if' statement!

ooRexx

ooRexx has a special singleton object called .nil that is used to indicate the absence of values in some situations (such as the default values returned from collection objects).

 
   if a[i] == .nil then say "Item" i "is missing"

Uninitialized ooRexx variables do not evaluate to .nil, but rather the character string name of the variable (all uppercase). The var() built-in function allows variable validity to be tested:

a=.array~of('A','B')
  i=3
  if a[i] == .nil then say "Item" i "of array A is missing"
  if \var("INPUT") then say "Variable INPUT is not assigned"
  if \var("var") then say "Variable" var "is not assigned"

Output:

Item 3 of array A is missing
Variable INPUT is not assigned
Variable VAR is not assigned

Oz

There is no explicit null in Oz.

Unbound variables

If an unbound variable is accessed, the current thread will be suspended:

declare
  X
in
  {Show X+2}  %% blocks

If you later assign a value to X in another thread, the original thread will resume and print the result of the addition. This is the basic building block of Oz' declarative concurrency.

Undefined values

Access to undefined values (like using an out-of-range array index or a non-existing record feature) will usually provoke an exception in Oz.

It is also possible to assign a unique "failed" value to a variable. Such a failed value encapsulates an exception. This can be useful in concurrent programming to propagate exceptions across thread boundaries.

declare
  X = {Value.failed dontTouchMe}
in
  {Wait X}  %% throws dontTouchMe

Sometimes algebraic data types like Haskell's Maybe are simulated using records.

declare
  X = just("Data")
in
  case X of nothing then skip
  [] just(Result) then {Show Result}
  end

PARI/GP

GP does not have good facilities for this, but this test suffices for most purposes:

foo!='foo

Pascal

See Delphi

PascalABC.NET

begin
  var a := new integer[](1,2,3);
  if Random(2) = 0 then
    a := nil;
  Print(a);
end.
Output:
nil
or
[1,2,3] 

Perl

In Perl, undef is a special scalar value, kind of like null in other languages. A scalar variable that has been declared but has not been assigned a value will be initialized to undef. (Array and hash variables are initialized to empty arrays or hashes.)

If strict mode is not on, you may start using a variable without declaring it; it will "spring" into existence, with value undef. In strict mode, you must declare a variable before using it. Indexing an array or hash with an index or key that does not exist, will return undef (however, this is not an indication that the index or key does not exist; rather, it could be that it does exist, and the value is undef itself). If warnings is on, most of the time, if you use the undef value in a calculation, it will produce a warning. undef is considered false in boolean contexts.

It is possible to use undef like most other scalar values: you can assign it to a variable (either by doing $var = undef; or undef($var);), return it from a function, assign it to an array element, assign it to a hash element, etc. When you do list assignment (i.e. assign a list to a list of variables on the left side), you can use undef to "skip" over some elements of the list that you don't want to keep.

You can check to see if a value is undef by using the defined operator:

print defined($x) ? 'Defined' : 'Undefined', ".\n";

From the above discussion, it should be clear that if defined returns false, it does not mean that the variable has not been set; rather, it could be that it was explicitly set to undef.

Starting in Perl 5.10, there is also a defined-or operator in Perl. For example:

say $number // "unknown";

prints $number if it is defined (even if it is false) or the string "unknown" otherwise.

Phix

There is a builtin NULL, however it is equivalent to the integer 0 and will trigger a type check if assigned to a variable declared as string or sequence. In most programs the zero-length string/sequence (""/{}) suffices, but if you want a variable that can be a string/sequence or NULL, but not other arbitrary integer/float values, use something like the following user-defined types:

type nullableString(object o)
    return string(o) or o=NULL
end type
nullableString s
s = "hello"
s = NULL
--s = 1 -- error
--s = {1,2,3} -- error
 
type nullableSequence(object o)
    return sequence(o) or o=NULL
end type
nullableSequence q
q = {1,2,3}
q = "string"    -- fine (strings are a subset of sequences)
q = NULL
--q = 1         -- error

See also Undefined_values

PHL

if (obj == null) printf("obj is null!\n");

PHP

There is a special value NULL. You can test for it using is_null() or !isset()

$x = NULL;
if (is_null($x))
  echo "\$x is null\n";

PicoLisp

New internal symbols are initialized with the value NIL. NIL is also the value for "false", so there is never really an "undefined value". 'not' is the predicate to check for NIL, but many other (typically flow control) functions can be used.

(if (not MyNewVariable)
   (handle value-is-NIL) )

or

(unless MyNewVariable
   (handle value-is-NIL) )

Pike

In Pike all variables are initialized to , regardless of their type. thus functions as a Null value for all types except integer.

is also used to indicate the absence of a key or object member.

to tell the difference between a value and absence of a key, zero_type() is used:

> mapping bar;
> bar;
Result: 0
> bar = ([ "foo":0 ]);
> bar->foo;
Result 0;
> zero_type(bar->foo);
Result: 0
> bar->baz;
Result: 0
> zero_type(bar->baz);
Result: 1

PL/I

declare x fixed decimal (10);
...
if ^valid(x) then signal error;

declare y picture 'A9XAAA9';
...
if ^valid(y) then signal error;

Comment:- In the picture specification, the content of variable y must consist of letters where the letter 'A' is given, digits or space where the digit '9' appears, and the letter X signfies that any character is acceptable.

PowerShell

In PowerShell the automatic variable $null represents a null value. Comparisons are not left/right symmetrical which means placing $null on the left side greatly assists when comparing to an array.

if ($null -eq $object) {
    ...
}

PureBasic

All variables that has not yet been given any other value will be initiated to #Null

If variable = #Null
  Debug "Variable has no value"
EndIf

Python

x = None
if x is None:
  print "x is None"
else:
  print "x is not None"

Output:

x is None

R

R has the special value NULL to represent a null object. You can test for it using the function is.null. Note that R also has a special value NA to represent missing or unknown values.

is.null(NULL)       # TRUE
is.null(123)        # FALSE
is.null(NA)         # FALSE
123==NULL           # Empty logical value, with a warning
foo <- function(){} # function that does nothing
foo()               # returns NULL

Racket

"null", or its literal form "'()", is used to denote empty lists and sometimes it is used as a generic null value.

-> null
'()
-> (null? null)
#t
-> (null? 3)
#f

But a value that is more used as a generic "nothing" value is "#f", false. Racket also has a void value, mostly the result of side-effect functions. (And an undefined value.)

Raku

(formerly Perl 6)

Translation of: Haskell

(as it were...)

In Raku you can name the concept of Nil, but it not considered an object, but rather the absence of an object, more of a "bottom" type. The closest analog in real objects is an empty list, but an empty list is considered defined, while Nil.defined always returns false. Nil is what you get if you try to read off the end of a list, and () is just very easy to read off the end of... :-)

If you try to put Nil into a container, you don't end up with a container that has Nil in it. Instead the container reverts to an uninitialized state that is consistent with the declared type. Hence, Raku has the notion of typed undefined values, that are real objects in the sense of "being there", but are generic in the sense of representing type information without being instantiated as a real object. We call these type objects since they can stand in for real objects when one reasons about the types of objects. So type objects fit into the type hierarchy just as normal objects do. In physics terms, think of them as "type charge carriers" that are there for bookkeeping between the "real" particles.

All type objects derive from Mu, the most-undefined type object, and the object most like "null" in many languages. All other object types derive from Mu, so it is like Object in other languages as well, except Mu also encompasses various objects that are not discrete, such as junctions. So Raku distinguishes Mu from Any, which is the type that functions the most like a discrete, mundane object.

Mostly the user doesn't have to think about it. All object containers behave like "Maybe" types in Haskell terms; they may either hold a valid value or a "nothing" of an appropriate type. Most containers default to an object of type Any so you don't accidentally send quantum superpositions (junctions) around in your program.

my $var;
say $var.WHAT;      # Any()
$var = 42;
say $var.WHAT;      # Int()
say $var.defined;   # True
$var = Nil;
say $var.WHAT;      # Any()
say $var.defined    # False

You can declare a variable of type Mu if you wish to propagate superpositional types:

my Mu $junction;
say $junction.WHAT;      # Mu()
$junction = 1 | 2 | 3;
say $junction.WHAT;      # Junction()

Or you can declare a more restricted type than Any

my Str $str;
say $str.WHAT;      # Str()
$str = "I am a string.";
say $str.WHAT;      # Str()
$str = 42;          # (fails)

But in the Raku view of reality, it's completely bogus to ask for a way "to see if an object is equivalent to the null object." The whole point of such a non-object object is that it doesn't exist, and can't participate in computations. If you think you mean the null object in Raku, you really mean some kind of generic object that is uninstantiated, and hence undefined. One of those is your "null object", except there are many of them, so you can't just check for equivalence. Use the defined predicate (or match on a subclass of your type that forces recognition of abstraction or definedness).

Raku also has Failure objects that, in addition to being undefined carriers of type, are also carriers of the reason for the value's undefinedness. We tend view them as lazily thrown exceptions, at least until you try to use them as defined values, in which case they're thrown for real.

Raven

This example is in need of improvement:

Add NULL handling with MySQL data.

NULL as $v
$v NULL =     # TRUE
$v NULL !=    # FALSE

1 NULL =      # FALSE
1.1 NULL =    # FALSE

NULL as $v2
$v2 $v =      # TRUE

REBOL

x: none

print ["x"  either none? x ["is"]["isn't"]  "none."]

Output:

x is none.

REBOL also has the concept of unset values, testable with get/any

unset? get/any 'some-var
unset? get 'some-var
Output:
true
** Script Error: some-var has no value
** Near: unset? get 'some-var

REXX

REXX can have variables with a null value.
With the   symbol   built-in function, it can be determined if a variable is defined (or not).
The   length   built-in function can be used to see what the length of the value of a defined variable.
A variable with a   null   value has a length of   0   (zero).

The   drop   statement can be used to "undefine" a REXX variable.

/*REXX program demonstrates null strings, and also undefined values. */

if symbol('ABC')=="VAR" then say 'variable ABC is defined, value='abc"<<<"
                        else say "variable ABC isn't defined."
xyz=47
if symbol('XYZ')=="VAR" then say 'variable XYZ is defined, value='xyz"<<<"
                        else say "variable XYZ isn't defined."
drop xyz
if symbol('XYZ')=="VAR" then say 'variable XYZ is defined, value='xyz"<<<"
                        else say "variable XYZ isn't defined."
cat=''
if symbol('CAT')=="VAR" then say 'variable CAT is defined, value='cat"<<<"
                        else say "variable CAT isn't defined."

output

variable ABC isn't defined.
variable XYZ is defined, value=47<<<
variable XYZ isn't defined.
variable CAT is defined, value=<<<

Ring

see isnull(5) + nl +        # print 0
isnull("hello") + nl +      # print 0
isnull([1,3,5]) + nl +      # print 0
isnull("") + nl +           # print 1
isnull("NULL")              # print 1

Ruby

The value when referring to the instance variable which isn't initialized is nil.

puts "@object is nil" if @object.nil?		# instance variable

puts "$object is nil" if $object.nil?		# global variable, too

# It recognizes as the local variable even if it isn't executed.
object = 1  if false
puts "object is nil" if object.nil?

# nil itself is an object:
puts nil.class  # => NilClass
Output:
@object is nil
$object is nil
object is nil
NilClass

Rust

// If an option may return null - or nothing - in Rust, it's wrapped
// in an Optional which may return either the type of object specified
// in <> or None. We can check this using .is_some() and .is_none() on
// the Option.

fn check_number(num: &Option<u8>) {
    if num.is_none() {
        println!("Number is: None");
    } else {
        println!("Number is: {}", num.unwrap());
    }
}

fn main() {
    let mut possible_number: Option<u8> = None;
    check_number(&possible_number);

    possible_number = Some(31);
    check_number(&possible_number);
}

S-lang

S-Lang uses NULL; it is the only object of type Null_Type:

variable foo = NULL;
print(foo);
if (foo == NULL)
   print(typeof(foo));
Output:
NULL
Null_Type

Scala

This blog post has a good explanations of the different types of null-like values.

scala> Nil
res0: scala.collection.immutable.Nil.type = List()

scala> Nil == List()
res1: Boolean = true

scala> Null
<console>:8: error: not found: value Null
              Null
              ^

scala> null
res3: Null = null

scala> None
res4: None.type = None

scala> Unit
res5: Unit.type = object scala.Unit

scala> val a = println()
a: Unit = ()

Scheme

(null? object)

Note: "null?" here tests whether a value is the empty list.

Sidef

The absence of a value is represented by nil

var undefined;         # initialized with an implicit nil
say undefined==nil;    # true
say defined(nil)       # false

However, nil is not an object, so we can't call methods on it. Alternatively, Sidef provides the null object:

var null_obj = null;        # initialize with a null value  
say null_obj.is_a(null);    # true
say defined(null_obj);      # true

Slate

Nil isNil = True.

Smalltalk

object isNil ifTrue: [ "true block" ]
             ifFalse: [ "false block" ].
nil isNil ifTrue: [ 'true!' displayNl ]. "output: true!"
foo isNil ifTrue: [ 'ouch' displayNl ].
x := (foo == nil).
x := foo isNil

notice that nil is the singleton instance of the UndefinedObject class; i.e. it is a first class object. Thus we can do:

foo := nil.
foo class. "-> UndefinedObject"
foo respondsTo: #'bar'. "asking if a message is implemented"

foo class compile:'fancyOperation ^ 123'.
foo fancyOperation "->123"

the last example being for demonstration only - it is not considered well behaved to add arbitrary code that way, except for framework support, such as encoding, decoding marshalling etc.)

Standard ML

Maybe the closest type of Standard ML would be the type option, which is defined like this in the standard library:

datatype 'a option = NONE | SOME of 'a
case v of NONE => "unbound value"
        | SOME _ => "bounded value"

Swift

Swift has Optional<T> type, where nil means a lack of value. T? is syntactic sugar for Optional<T>.

let maybeInt: Int? = nil

To just check if variable is nil, you can use == operator.

if maybeInt == nil {
  print("variable is nil")
} else {
  print("variable has some value")
}

Usually you want to access the value after checking if it's nil. To do that you use if let

if let certainlyInt = maybeInt {
  print("variable has value \(certainlyInt)")
} else {
  print("variable is nil")
}

Tailspin

Tailspin does not have a null value, but a transform is allowed to produce nothing at all, in which case that chain of computation simply stops. A templates transform can explicitly label cases as producing nothing by !VOID but also input values for which there is no matching branch will produce nothing. If you need computation to continue even in the event of nothing being produced, you can wrap the transform in an array/list to get an empty list.

templates mightBeNothing
  when <=0> do !VOID
  when <=1> do 'something' !
end mightBeNothing

1 -> mightBeNothing -> 'Produced $;. ' -> !OUT::write

0 -> mightBeNothing -> 'Won''t execute this' -> !OUT::write
2 -> mightBeNothing -> 'Won''t execute this' -> !OUT::write

// capture the transform in a list to continue computation when no result is emitted
[1 -> mightBeNothing] -> \(
  when <=[]> 'Produced nothing. ' !
  otherwise 'Produced $(1);. ' !
\) -> !OUT::write

[0 -> mightBeNothing] -> \(
  when <=[]> 'Produced nothing. ' !
  otherwise 'Produced $(1);. ' !
\) -> !OUT::write

[2 -> mightBeNothing] -> \(
  when <=[]> 'Produced nothing. ' !
  otherwise 'Produced $(1);. ' !
\) -> !OUT::write
Output:
Produced something. Produced something. Produced nothing. Produced nothing. 

It is an error to try to assign nothing to a symbol or field.

// throws an error
def nothing: 0 -> mightBeNothing;

// throws an error
{ nothing: 0 -> mightBeNothing }

// OK, simply results in the empty structure without a field called 'nothing'
{ 0 -> mightBeNothing -> (nothing: $) }

Tcl

In Tcl, where every value is a string, there is no out-of band value corresponding to NULL. In many cases, using the empty string is sufficient:

if {$value eq ""} ...

A stricter approximation to NULL can be had with non-existing variables or elements of a dict or array:

if {![info exist nullvar]} ...
if {![info exists arr(nullval)]} ...
if {![dict exists $dic nullval]} ...

Note that lists do not support anything like nulls, since they are strictly sequences of values.

TXR

Translation of: Common Lisp

TXR Lisp has a nil symbol which serves as the empty list and Boolean false, like Common Lisp and similar dialects. It is a very important symbol which plays a central role.

Variable definitions, global and local, without initial value expressions take on the value nil. Elements of newly created vectors and structure slots are nil by default. Optional function parameters that don't receive arguments are given nil arguments by default. Many functions which search for something use nil for indicating not found.

Object-oriented programming in TXR Lisp is done with structures, which do not provide a CLOS-like object system. The nil object isn't a structure, and so it cannot take slot references, or method invocation. Only structures have methods, which means that it's not possible to define a null object method m such that obj.(m) can be invoked if obj is an expression evaluating to nil.

The Gang-of-Five Null Object Pattern can be employed: defining struct types that behave like null. This null object doesn't have to be related by inheritance to the structs in conjunction with which it is used.

(defstruct null-widget ()
  (:method popularity (me) 0))

(defvarl null-widget (new null-widget))

(defstruct real-widget ()
  pop
  (:method popularity (me) me.pop))

In situations when a null object would be used simply to provide safe treatment of null without the verbosity of checks for its presence, and a default value of nil is acceptable, TXR Lisp has null-safe slot and method access: obj.?slot, obj.?(method arg).

The expression w.?(popularity) will evaluate to nil if w is nil, being equivalent to (if w w.(popularity)), except that w is evaluated only once.

Like in Common Lisp, nil an instance of the type null, which is a unit type having only that instance. And nil is also the name of the bottom type of the type system: the nil type is the subtype of every type, including itself.

Like in Common Lisp, the expression (or a b c ...) evaluates the arguments from left to right, returning the value of the leftmost one which does not evaluate to nil.

;; Find widget in one of three places in order
(defun find-widget (name)
  (or [%widget-hash% name]
      (lookup-local-widget name)
      (lookup-global-widget name)))

Ursa

# the type at declaration doesn't matter
decl int x

set x null
if (= x null)
    out "x is null" endl console
else
    out "x is not null" endl console
end if

VBA

Public Sub Main()
    Dim c As VBA.Collection
     
    ' initial state: Nothing
    Debug.Print c Is Nothing
     
    ' create an instance
    Set c = New VBA.Collection
    Debug.Print Not c Is Nothing
     
    ' release the instance
    Set c = Nothing
    Debug.Print c Is Nothing
End Sub

Visual Basic

Works with: VB6

Null by the definition of this task is called "Nothing" in VB6:

Public Sub Main()
Dim c As VBA.Collection

' initial state: Nothing
Debug.Assert c Is Nothing

' create an instance
Set c = New VBA.Collection
Debug.Assert Not c Is Nothing

' release the instance
Set c = Nothing
Debug.Assert c Is Nothing

End Sub

The Null keyword has a different meaning in VB6: it's one of the states that a Variant type variable can be in. Null means that a Variant doesn't hold valid (i.e.: defined) data.

Public Sub Main()
Dim v As Variant

' initial state: Empty
Debug.Assert IsEmpty(v)
Debug.Assert VarType(v) = vbEmpty

v = 1&
Debug.Assert VarType(v) = vbLong

' assigning the Null state
v = Null
' checking for Null state
Debug.Assert IsNull(v)
Debug.Assert VarType(v) = vbNull

End Sub

V (Vlang)

V (Vlang) does not have nor normally allow null or nil (also called "the billion dollar mistake"):

1) All primitive types/variables have default values.

2) For arrays and structs, default values are automatically assigned.

3) Usage of voidptr or nil, only done in unsafe, and is usually only for the purposes of interoperability with other languages.

// Null or nil not used, default values for various primitive types/variables instead:

a_string :=''
a_bool := false
an_int := 0
[3]string{} // ['', '', '']
[3]bool{} // [false, false, false]
[3]int{} // [0, 0, 0]

Wart

The null value nil is also the only false value.

(not nil)

Wren

In Wren, it is not technically possible for a variable to have no value at all.

If you define a variable without giving it a value, then it is given the special value null which is the only instance of the Null class and also a keyword.

Similarly, if you define a function but don't give it a return value, then it returns null. In this particular respect, null is similar to void in C.

In boolean expressions, null is considered to be false whereas (apart from false itself) all other values (even 0) are considered to be true. For consistency the Null class therefore overrides the ! operator, which it inherits from the Object class, so that !null returns true.

It is always easy to test for nullness either by querying a variable's type or checking the value of a boolean expression involving a potentially null variable.

// Declare a variable without giving it an explicit value.
var s

// We can now check for nullness in one of these ways.
System.print(s)
System.print(s == null)
System.print(s is Null)
System.print(s.type == Null)

// Similarly, if we define this function without giving it a return value.
var f = Fn.new {
    System.print("I'm a function with no explicit return value.")
}

// And now call it.
var g = f.call()
// We find that the return value is null.
System.print(g)
Output:
null
true
true
true
I'm a function with no explicit return value.
null

zkl

In zkl, there isn't a C like 0/NULL, a value that, if referenced, causes bad things to happen. There is an Object, Void, that is used as generic NULL like thing but it is just another object.

if(Void == n) ...
return(Void)

Z80 Assembly

Translation of: ARM Assembly

Even though this doesn't really apply to assembly, as there is no "null pointer" per se (that is, all data is a number), it would be interesting to demystify what NULL really is. At the lowest level, the null pointer is just a pointer to a memory location that is of no use to the programmer. The Z80 does not segfault, so any memory address we read is fair game. So without the hardware enforcing NULL, we need to pick an address we don't mind losing. Typically, using &0000 to equal C's NULL is acceptable, as address &0000 on any randomly chosen Z80-based hardware is typically a jump to the kernel's entry point (or the main program's entry point, depending on the implementation.)

ld a,(&0000)   ;dereference the null pointer as a uint8
ld hl,(&0000)  ;dereference the null pointer as a uint16

Typically, you would get 0xC3 when dereferencing as an 8-bit value and 0xC3nn when dereferencing as a 16-bit value, where nn is the low byte of the address of the aforementioned entry point. Neither of these are particularly useful, so having &0000 as the null pointer is perfectly fine. Although there are technically other options, most CPUs can compare with zero more efficiently than most other numbers, and the Z80 is no exception. Of course, the Z80 will not check if a pointer is NULL for you, so you have to do it yourself:

LD HL,myPointers
;there is no LD BC,(HL) so we have to do this:
LD c,(hl)
inc hl
LD b,(hl)
;and compare to null
LD a,b
or c    ;compare BC to zero
JR z,isNull
Cookies help us deliver our services. By using our services, you agree to our use of cookies.