This happens about 1 in 5 times i exit from my game. I'm getting an assertion from the microsoft library (delete operator method in the dbgdel.cpp file) and it's at this line:

_ASSERTE(_BLOCK_TYPE_IS_VALID(pHead->nBlockUse));

Anyone got a clue what it means! all i'm doing is deleting some memory, as in:

delete car;

Neil.

hello,
error is _CrtIsValidHeapPointer(pUserData), line 1132 of dbgheap.c
/*

• If this ASSERT fails, a bad pointer has been passed in. It may be

• totally bogus, or it may have been allocated from another heap.

• The pointer MUST come from the 'local' heap.

*/
_ASSERTE(_CrtIsValidHeapPointer(pUserData));

the line that my debugger stops me in is in the file dbgdel.cpp (delete operator) on the ASSERT line below:

void operator delete(
void *pUserData
)
{
_CrtMemBlockHeader * pHead;

RTCCALLBACK(_RTC_Free_hook, (pUserData, 0));

if (pUserData == NULL)
return;

_mlock(_HEAP_LOCK); /* block other threads */
__TRY

/* get a pointer to memory block header */
pHead = pHdr(pUserData);

/* verify block type */
_ASSERTE(_BLOCK_TYPE_IS_VALID(pHead->nBlockUse));

_free_dbg( pUserData, pHead->nBlockUse );

__FINALLY
_munlock(_HEAP_LOCK); /* release other threads */
__END_TRY_FINALLY

return;
}

this only happens now and again. when i debug the code where it fails the object i'm deleting is perfectly valid and i can inspect all its data.

i don't think so as when i debug the code all data is fine and the code isn't doing anything different i can see when it fails or when it works.

however, i've moved the delete statement (deleting some memory created inside a struct) to the destructor of the struct and it seems to behaving itself a bit better. though time will tell

Neil.

Same error I got.
You are deleting a peice of memory that was allocated in another heap (in another DLL for example).

A double delete (dangling pointer), a non-initialized pointer, or a pointer from a different heap will cause this assert. A NULL pointer does not cause it, as X-G conjectured, because deleting NULL pointers is safe and well-defined.

Actually, deleting a NULL pointer under MSVC 6 will cause an assert. I've run into it, kind of annoying. Not that it's bad, or will crash in Release mode, but it certainly asserts.

I'm also wondering about this 'delete 0' causes an assert. Never seen that, I'm using MSVC 6.0 7.0 7.1 for years now.

I'm confused. You can't write delete 0, or delete NULL. NULL (and '0') isn't a pointer. So I'm just going to assume you mean variable containing a null pointer.

If this:

int *temp = new int;
delete temp;
delete temp;

is what you're talking about, yes, it crashes (or gives an unhandled exception in debug mode) when using MSVC6.

How good do you expect a compiler to be that's made before the specification is standardized?

Yes, I guess I just misinterpreted what you were saying.

Man, I have bad memory I had always recalled getting asserts on NULLs. A test program under MSVC 6 shows nothing. Ok, my bad.