Thread: Static member func clone?

Hi, i ran into a really weird problem:

Take the following code:

Code:

class CExample{
        static CMemoryBuffer<CPlayerDescription>* Allocator()
        {
	static CMemoryBuffer<CPlayerDescription>* instance = new CMemoryBuffer<CPlayerDescription>();
	return instance;
        }
};

Now, i have another class within 2 member functions. Both functions do a call to CExample::Allocator to retrieve the allocator. However, weird enough they both got their own instances.

Is this normal behaviour? because the method is static, shouldn't it return the same instance?

Can you post a small but complete and compilable example that shows the problem?

Hi daved:

This is what i do: (Couldnt copy/paste the original code but the structure is exactly the same

Code:

class CShared{
	void doSomething() { } 
};

class CFoo{
public:
	static CShared* get_SharedInstance(){
		static CShared* pObjShared = new CShared();
		return pObjShared;
	}
};

class CBar{
	void Method1{
		CShared* pObjShared = CFoo::get_SharedInstance();
		std::cout << pObjShared << std::endl;
	}
	void Method2{
		CShared* pObjShared = CFoo::get_SharedInstance();
		std::cout << pObjShared << std::endl;
	}
};

Method 1 prints: address1
Method 2 prints: address2

The application is multithreaded, better said, there is little change that the same thread who called method 1, calls method 2 on the same instance. Might this have something to do with it?

I use singleton classes alot in my application ( yeh i know ugly ) so is surely hope that this problem doesn't exists in those classes.

I solved it temporary by using a funvtion returning that object which works fine

I was kind of hoping for something that I could plop into my compiler, build it and run it and see the same results. I was able to make some changes to that code and get it to compile, but it gave the same value for both methods (VC++ 7.1).

Part of reducing the problem to a complete and compilable example is that it might help you narrow down the issue. The multithreaded-ness of the app might have something to do with it, I'm not sure. If you can make a small sample that works in a singlethreaded system but not a multithreaded system, that would be a good indicator. Regardless, unless somebody knows from experience what the issue is, finding the smaller example might be the best course of action.

>Is this normal behaviour? because the method is static, shouldn't it return the same instance?

The pointer itself is static but the address assigned to it is not. Try statically allocating the CMemoryBuffer object itself instead of using new and see what happens.

But a static pointer should only be initialized once, so new should only be called once.

Another option is to put a breakpoint on that line and see from where and when it gets called.