CRT reports memory leaks

Just started some testing of my resource manager and the CRT is reporting leaks but they are not being dumped at program termination. My program exits with code 0 and nothing is dumped.

So why would the CRT report memory leaks and yet none of them are dumped?

If I create the skysphere and sun CRT will report a mem leak on exit but no blocks are reported or dumped. If I only create the skysphere or the sun but not both, CRT does not report anything and as well no blocks are dumped.

Quite odd.

CResource.h

Code:

class IResource
{
  public:
    virtual IResource* Get()=0;
    virtual void* GetBuffer() const=0;
    virtual UINT GetSize() const=0;
    virtual DWORD GetID() const=0;
    virtual void SetID(const DWORD ID) const=0;
    
};

class CResource:public IResource
{
  private:
    CResource(const CResource &res);
    CResource &operator=(const CResource &res);
    
  protected:
    //Size of file (and hence this resource)
    UINT m_uFilesize;
    
    //Resource unique ID - must be unique or system will fail
    mutable DWORD m_dwID;
            
    //Resource data
    BYTE *m_pBuffer;
    
  public:
    CResource(std::string Filename);
    virtual ~CResource()
    {
      //::MessageBox(0,"Deleting resource",0,0);
      delete [] m_pBuffer;
    }
    ...
    ...

};

CResMgr.h

Code:

...
...
class CResource;

//Typedef for cache list and map iterators
typedef std::list<CResource *>::iterator CacheList_Iter;
typedef std::map<DWORD,CacheList_Iter>::iterator CacheMap_Iter;



class CResMgr
{
  private:
    //Hide these two to prevent accidental copying
    CResMgr(const CResMgr &); 
    CResMgr &operator=(const CResMgr &);
  
        
    //Cache stats
    DWORD m_dwDesiredSize;
    DWORD m_dwCurSize;
    DWORD m_dwNumHits;
    DWORD m_dwNumMisses;
    
    DWORD m_dwMaxSize;
            
  protected:
    
    //STL map and list
    std::list<CResource *> CacheList;
    std::map<DWORD, CacheList_Iter> CacheMap;
    ...
    ...
};

CResMgr.cpp

Code:

...
...

CResMgr::~CResMgr()
{
  //Delete the list
  CacheList_Iter List_Iter=CacheList.begin();
  
  for (UINT i=0;i<static_cast<UINT>(CacheList.size());i++);
  {
    delete *List_Iter;
    List_Iter++;
  
  }
  
  CacheList.clear();
  CacheMap.clear();
  
}

Am I leaking in the destructor? I think I've posted all of the code needed to answer this.

> for (UINT i=0;i<static_cast<UINT>(CacheList.size());i++);
There's a classic trailing ; here

OMG. Let me check my original code and see if that is there. If it is I feel more than stupid.

Hehe.

Thanks for pointing it out.

Is it the newest VC++ or GCC that warns about this stuff? I know at least one of them can do it. (And if you really need an empty loop, use empty braces.)

My copy of MinGW doesn't seem to do it (gcc/g++ version 3.4.5). I wonder if I'll ever see an update to MinGW again in my lifetime. They sure are taking their time about it.

Originally Posted by Frobozz

My copy of MinGW doesn't seem to do it (gcc/g++ version 3.4.5). I wonder if I'll ever see an update to MinGW again in my lifetime. They sure are taking their time about it.

they'll port gcc-4.2 some time...

3.4.5 is hardly the newest version

Warnings for GCC, from http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html

-Wextra [or -W]
- (C++ only) An empty body occurs in a `while' or `for' statement with no whitespacing before the semicolon. This warning can be independently controlled by -Wempty-body.

-Wempty-body
Warn if an empty body occurs in an `if', `else' or `do while' statement. Additionally, in C++, warn when an empty body occurs in a `while' or `for' statement with no whitespacing before the semicolon. This warning is also enabled by -Wextra.

Originally Posted by CornedBee

3.4.5 is hardly the newest version

Which is sad because I'm using the "candidate" install for MinGW. I wish they'd stop spending so much time updating their stupid site layout (confusing me each time in the process) and spend more time updating their software package.