Requesting Affirmation [pointer 2 custom class]

[Deo]

Boy, that dictionary is paying off

Anyways.

Heres my problem:

I have a custom class I created called CTexture as well as a custom class called CSprite. What I need is for multiple CSprite classes to point to a SINGLE CTexture class.

Heres my two classes.

class CTexture
{
  public:
    CTexture();
    ~CTexture();
    LPDIRECT3DTEXTURE9 GetTexture();
    int GetWidth();
    int GetHeight();		
    bool LoadTexture(char *filename, int width, int height, LPDIRECT3DDEVICE9 device);

  private:
  LPDIRECT3DTEXTURE9 m_texture;
  int m_width;
  int m_height;
};

struct SFrame
{
  int frameNum;
  RECT frameBoundary;
  D3DXVECTOR3 center;	
  SFrame *nextFrame;
};

class CSprite
{
  public:
    CSprite();
    ~CSprite();
    bool SetTexture(CTexture &texture);
    void NewFrame(int frameWidth, int frameHeight, int startPixelX, int startPixelY);
    bool Draw(int frameNum,  D3DXVECTOR3 position, LPDIRECT3DDEVICE9 device);

  private:
    CTexture *m_Texture;
    SFrame *m_FirstFrame;
    LPD3DXSPRITE m_Sprite;
    int m_TotalFrames;
};

bool CSprite::SetTexture(CTexture &texture)
{
    if(!texture.GetTexture())
       return false;
    
    m_Texture = &texture;
	return true;
}

I was just wondering if my line in RED is correct or not. I'm pretty sure it is and by cycling through the debugger abit it appears to do what I want it to do. However, since this wasn't my initial solution I just wanted to get a second opinion if this is proper or not.

My initial solution was:

m_Texture = texture;

But it was complaining about "=" because I didn't overload the "=" for CTexture class. If I created a "=" function and copied variables from texture to m_Texture that would create a copy and not a pointer right?

Because elsewhere in my code I have:

SFrame *m_FirstFrame;
SFrame *temp;
temp = m_FirstFrame;

And it works fine.. seems like a similar circumstance except temp & m_FirstFrame are referring to a STRUCT while in the first case I had pointers to a CLASS.. Why does the first example require a & but the second case does not complain about a need for overload "=" operator.

Is this because a class has private while everything in a struct is public?

Thnx for any clarification.

[Mortissus]

And it works fine.. seems like a similar circumstance except temp & m_FirstFrame are referring to a STRUCT while in the first case I had pointers to a CLASS.. Why does the first example require a & but the second case does not complain about a need for overload "=" operator.

In the second case, both variables are of the same type: SFrame *

In the first case, the types are distinct ( CTexture& != CLTexture *). Your code seems ok, but I would make CSprite::m_Texture a reference, and not a pointer. If you do this, you will need to initialize the variable at the constructor:

CSprite::CSprite(CTexture& t)
:m_Texture(t)
{
   //code
}

Do not forget to change the m_Texture type to & instead of *.

[Deo]

Thnx for the info...

Your code seems ok, but I would make CSprite::m_Texture a reference, and not a pointer.

Ok, your the boss... will do... just one quick question. If I were to make this variable a reference and not a pointer what would happen when my CSprite variable goes out of scope and the destructor is called. Since my m_texture is a reference wouldn't that then call the destructor for my CTexture class?

Just wanted to make sure it wouldn't as I will have one CTexture class shared among multiple CSprite classes.

[Mortissus]

Yes, it will. I am sorry I have not seen this before. You are completely right. Change the references to pointers. I would sugest still use the constructor to set the texture, if this is not possible use the method as before. Again, I apologize for my mistake telling you to use references.

[Daved]

No, if m_Texture is a reference the CTexture destructor will NOT be called.

The decision of whether to make it a reference or a pointer should be based on whether you ever need it to be null, or whether you ever need to change the m_Texture for a specific CSprite instance. If neither will happen, a reference is fine. Otherwise, your original code that sets the pointer is fine.

I assume you are managing the memory containing the CTexture instances somewhere else. If so, either choice should work, just don't delete the CTextures until the CSprites are no longer used.

[Mortissus]

Uhm... I thought that would be destroyed. Thanks for the correction! . What a shame again!