If Me Is Nothing

djork

Someone please tell me I'm not batshit-flipping-insane to think that "Me" in VB or "this" in C# can never ever possibly be null.

Thanks.

Nandurius

It should never be null, in a properly designed application. I'm pretty sure it can end up being null through various bugs though.

I don't know how VB compiles OO code, but a C++ function has a "this" parameter that references the object data. A function call such as "object.function(arg1);" would call objectType_function_param1(object, arg1);

Now, if object is set to null, the "this" argument passed to the function implementation is null. What one *should* be doing is find where a function is called on an object that has already been destroyed, rather than trying to catch those calls and ignore them. This is a band-aid patch for a bug in another place.

Devi

I'm fairly sure that in C# this situation is impossible, as far as I know the system should throw a null pointer exception before even executing any member function with an empty reference. C# won't allow you to work with raw pointers to objects, so bypassing the reference system doesn't really seem to be an option either.

I have seen people checking for this==NULL in C++ before though, some lunatic thought that:

bool MyObjectTemplate::InheritsFrom(ClassID classID)

{

if (this==NULL) return false;

if (mClassID==classID) return true;

return ParentClass::InheritsFrom(classID);

}

Would be a perfect way to fix some of the crashes he'd been having. He also had to remove the virtual declaration to make it work. Sigh.

djork

My original post was written hastily. I was leaving work.

So, yeah, I understand how it can be possible in many languages that manage OOP by implicitly taking this as an argument (C++, Lua, Python sort of). I just can't imagine how, in a .NET language, this could ever be null when NullReferenceExceptions are thrown before an instance method is called on a null object.

I'm 99.99% sure that it's impossible, but how can I positively prove it?

asuffield

@djork said:

I'm 99.99% sure that it's impossible, but how can I positively prove it?

You can't prove that an MS product doesn't contain stupid bugs.

@djork said:

I'm 99.99% sure that it's impossible, but how can I positively prove it?

@The Microsoft® Visual Basic® Language Specification, Version 8.0, 11.4.3 said:

• The keyword Me represents the instance of the type containing the method or property accessor being executed.

@Same, 2.4.7 said:

Nothing is a special literal; it does not have a type and is convertible to all types in the type system, including type parameters. When converted to a particular type, it is the equivalent of the default value of that type.

@Same, 7.1 said:

A [i]null reference[/i] is a reference that refers to nothing; it is not possible to do anything with a null reference except assign it. <...> For a variable of a reference type, the default value is a null reference.

QED?

djork

@Spectre said:

@djork said:
I'm 99.99% sure that it's impossible, but how can I positively prove it?

@The Microsoft® Visual Basic® Language Specification, Version 8.0, 11.4.3 said:
• The keyword Me represents the instance of the type containing the method or property accessor being executed.

So, with that in mind, in VB .NET I can safely assume that all of the "If Me Is Nothing" checks can be nuked. If the app crashes willy-nilly then I'll let Microsoft know.

according to the language it will never be null. So short of a compiler error or something you will be safe.

XIU

579: public Object CreateInstance(String typeName,
580: bool ignoreCase,
581: BindingFlags bindingAttr,
582: Binder binder,
583: Object[] args,
584: CultureInfo culture,
585: Object[] activationAttributes)
586: {
587: // jit does not check for that, so we should do it ...
588: if (this == null)
589: throw new NullReferenceException();

http://www.dotnet247.com/247reference/System/Reflection/Assembly/__rotor

Checked with the released sources and reflector...