Allegro.cc - Online Community

Allegro.cc Forums » Programming Questions » 'NULL' was not declared in this scope

This thread is locked; no one can reply to it. rss feed Print
 1   2 
'NULL' was not declared in this scope
BAF
Member #2,981
December 2002
avatar

You have null be a keyword that means NULL (a value which could never be held by any variable). Doesn't have a literal definition, just a symbolic one to the compiler.

Arthur Kalliokoski
Second in Command
February 2005
avatar

Quote:

(a value which could never be held by any variable)

And what would that value be, without going to 96 bit ints or whatever? Even if it's a #define, any equivalent sequence of bits could be in a regular integer variable.

Nothing matters and what if it did

Thomas Fjellstrom
Member #476
June 2000
avatar

Quote:

And what would that value be, without going to 96 bit ints or whatever? Even if it's a #define, any equivalent sequence of bits could be in a regular integer variable.

It would have to be a compile time thing. Its hard to do if you never plan for this sort of feature ahead of time.

Perl has a special "undef" flag on scalars, its not 0, its not any kind of null or "" (empty string), its special :) and perl complains loudly if you use undefined variables (if you tell it to of course, but I always do, it finds so many bugs, and has trained me to not use undefined variables).

--
Thomas Fjellstrom - [website] - [email] - [Allegro Wiki] - [Allegro TODO]
"If you can't think of a better solution, don't try to make a better solution." -- weapon_S
"Goto is the buldozer of coding. Sometimes, the buldozer is just the right tool for the job. Not often, but sometimes." -- LordBob

X-G
Member #856
December 2000
avatar

That would not work for C++. It works in dynamically typed languages because you can have a Null type, which can only be a single value -- null (or None, or Nothing, or whatever you like). That way, you can assign the null constant to any variable at any point. Python and PHP both work this way. C++ is not dynamically typed, so this solution is out.

As far as I know, Perl's undef construct is different; it directly manipulates the relevant symbol table, removing the identifier associated with a particular variable. PHP can do this too, I believe. Obviously, this will not work for a statically typed language like C++.

--
Since 2008-Jun-18, democracy in Sweden is dead. | 悪霊退散!悪霊退散!怨霊、物の怪、困った時は ドーマン!セーマン!ドーマン!セーマン! 直ぐに呼びましょう陰陽師レッツゴー!

Thomas Fjellstrom
Member #476
June 2000
avatar

Quote:

That would not work for C++.

Obviously not, it was just for comparison's sake. As I mentioned, anything C++ could do would have to be magic compiler syntax that ends up being mostly useless since you wouldn't be able to check the undef/nullity state at runtime with any sort of accuracy.

--
Thomas Fjellstrom - [website] - [email] - [Allegro Wiki] - [Allegro TODO]
"If you can't think of a better solution, don't try to make a better solution." -- weapon_S
"Goto is the buldozer of coding. Sometimes, the buldozer is just the right tool for the job. Not often, but sometimes." -- LordBob

Tobias Dammers
Member #2,604
August 2002
avatar

It does not have anything to do with dynamic vs. static typing, nor with compiled vs. interpreted. For example, C#, being compiled (at least half-way), and with basically static typing, DOES have a null keyword, and C#'s null does not compare equal to zero.
The definition of a language's syntax does not mandate a certain bit representation in memory; in fact, even the existing C++ standard, as far as I know, doesn't say anything about the way numbers are stored in bytes. Nor does it mandate that 0 in a pointer context is stored exactly like in an integer context; it does mandate, though, that using 0 in a pointer context refers to the NULL pointer, and that when converting between pointer and integer types, 0 (the NULL pointer) is equivalent to 0 (the number zero).

---
Me make music: Triofobie
---
"We need Tobias and his awesome trombone, too." - Johan Halmén

X-G
Member #856
December 2000
avatar

That doesn't change the fact that C++ does not have anything like variable type flags; a long int may be 32 bit, say, and nothing more or nothing less, and each of those 32 bits are accessible. There is no "metadata" to say "this is null". In other words, null is indistinguishable from a particular value that datatype can contain, such as the number zero.

--
Since 2008-Jun-18, democracy in Sweden is dead. | 悪霊退散!悪霊退散!怨霊、物の怪、困った時は ドーマン!セーマン!ドーマン!セーマン! 直ぐに呼びましょう陰陽師レッツゴー!

Tobias Dammers
Member #2,604
August 2002
avatar

An int cannot be null; a pointer can, though, and C++ is capable of adding type information to those.
The whole discussion is nonsense, though, because C++ with different pointer semantics is, well, maybe D or C# or something.

---
Me make music: Triofobie
---
"We need Tobias and his awesome trombone, too." - Johan Halmén

bamccaig
Member #7,536
July 2006
avatar

// nullcmp.cs
using System;

namespace NullCmp
{
    public class Program
    {
        private static int Main(string[] args)
        {
            System.Console.WriteLine("(null==0)={0}", (null == 0));

            return(0);
        }
    }
}

prompt>csc nullcmp.cs
Microsoft (R) Visual C# 2005 Compiler version 8.00.50727.1434
for Microsoft (R) Windows (R) 2005 Framework version 2.0.50727
Copyright (C) Microsoft Corporation 2001-2005. All rights reserved.

test.cs(10,46): warning CS0472: The result of the expression is always 'false'
        since a value of type 'int' is never equal to 'null' of type 'int?'
prompt>test
(null==0)=False

Interestingly, apparently "null" is of type "int?". I'm curious how it is stored under the surface.

1// nullcmp2.cs
2using System;
3 
4namespace NullCmp
5{
6 public class Program
7 {
8 private static unsafe int Main(string[] args)
9 {
10 int* i = null;
11 
12 System.Console.WriteLine("({0}==null)={1}", (int)i, (i == null));
13 
14 i = (int*)0;
15 
16 System.Console.WriteLine("({0}==null)={1}", (int)i, (i == null));
17 
18 return(0);
19 }
20 }
21}

prompt>csc nullcmp2.cs /unsafe
Microsoft (R) Visual C# 2005 Compiler version 8.00.50727.1434
for Microsoft (R) Windows (R) 2005 Framework version 2.0.50727
Copyright (C) Microsoft Corporation 2001-2005. All rights reserved.

prompt>test
(0==null)=True
(0==null)=True

I may have done something wrong or perhaps the results don't represent the actual implementation, but it seems possible that C# is storing null as 0. There's a good chance I'm overlooking something or lack the understanding to see the difference. I don't actually know how pointers or unsafe code works in C# yet so this is all just a guess. :P

Arthur Kalliokoski
Second in Command
February 2005
avatar

Quote:

An int cannot be null

Third time I've posted this...
NULL == NIL == NADA == NOTHING == 0 == ZILCH.

Nothing matters and what if it did

Tobias Dammers
Member #2,604
August 2002
avatar

Quote:

NULL == NIL == NADA == NOTHING == 0 == ZILCH.

Not true. Conceptually, Null is something very different from zero, and most languages (but not C / C++) treat it as such - even though in many of them, NULL (or the language equivalent) compares equal to zero due to implicit conversions.
PHP even distinguishes between Null and undefined; although the difference is somewhat esoteric, it does make sense in the PHP universe.

Quote:

Interestingly, apparently "null" is of type "int?". I'm curious how it is stored under the surface.

In C#, null is of type Null (a class that can only ever carry the null value). Since you're comparing it to an int, though, the compiler applies an implicit conversion, using a type that can both be assigned a null value AND be equality-compared to an int. The best match for these constraints is a nullable int (int?), so that's what the null in your example is treated as. If you were comparing it to a bool, you'd get a bool?. For a string (which is a reference type and thus nullable by default), you'd get a string (not a string?).

---
Me make music: Triofobie
---
"We need Tobias and his awesome trombone, too." - Johan Halmén

bamccaig
Member #7,536
July 2006
avatar

According to Pointers in C#, if you cast a pointer to an int you get the memory address though. In theory, null still could be a special circumstance, but to me at least, it seems more likely that 0 is the actual "address"/value stored in the pointer. Abstracting it doesn't really change the underlying implementation.

Tobias Dammers
Member #2,604
August 2002
avatar

In fact, using unsafe code, it is. In unsafe mode, C# allows pointers which work pretty much like those in C++, and they can be cast to and from integers - just like in C++. And in that case, a null pointer (as opposed to a null reference!) IS stored as integer zero, and converts to 0 when cast to int.
For reference types, this doesn't work, because C# doesn't allow pointers to reference-typed objects.

---
Me make music: Triofobie
---
"We need Tobias and his awesome trombone, too." - Johan Halmén

bamccaig
Member #7,536
July 2006
avatar

A pointer is essentially an integer. An integer can store any combination of bits and, therefore, there is no "special" value (0, excluded) for NULL to be without abstracting pointers and adding metadata to them. There would be absolutely no advantage to that. If anything, it would slow the process down and consume more memory than necessary.

References are essentially pointers that are managed by the language/compiler. I'm going to assume that more than likely, null is still represented as 0 for references (that can have a null reference) on an assembler (or intermediate) level and the whole interaction is abstracted from the programmer so it seems more complex than it really is. There may be more to references than I think, in which case I could be mistaken, but I can't think of any reason for them to be more complex than that.

 1   2 


Go to: