Re: Stupid *ptr problems
Yes, its plain old regular C. The entire program is old and is
regular C. I happens and you have to deal with the card you are
given.
GlobalAlloc is used through out the entire program and all its dlls.
There is no need to change or mix methods as that will just confuse
any future developers.
Yes, GlobalLock and unlock are dead, again they are in the code and no
need to remove them.
Removing the equal from teh for() kills the buffer overflow.. good
catch but not the problem! LOL Actually that for loop is inside an
else statement, if nothing was returned from the ODBC query then the
for loop would never get processed anyway.
Once Fetch returns a result, an initial block of memory is allocated
to the size of QUERYLOG and each element within the structure receives
a _strdup() from the corresponding field from the query.
Since I have no idea how many results will be returned (could be 0, 1
or 100,000) I expand the memory block with GlobalRealloc() from the
2nd Fetch onward.
essentially I should have a memory array consisting of:
[QUERYLOG][QUERYLOG][QUERYLOG][QUERYLOG] etc.
I think pQueryLog = [QUERYLOG][QUERYLOG][QUERYLOG][QUERYLOG]
should give me the first element while pQueryLog++ should give me the
2nd.
Or pQueryLog = [QUERYLOG][QUERYLOG][QUERYLOG][QUERYLOG]
can also be referenced as pQueryLog[nCnt].xxx. If nCnt = 3 then I
should get the fourth element.
Am I missing something here?
<eckhardt@satorlaser.com> wrote:
Howdy wrote:
I am updating an old program and it uses GlobalAlloc() I am expanding
a block of memory to write structure of pointers to what should end up
as an array of pointers.
I don't see anything particular that justifies the use of GlobalAlloc().
malloc() or for C++ std::vector should work just fine.
typedef struct {
LPSTR pszStartDate;
LPSTR pszStartTime;
LPSTR pszEndDate;
LPSTR pszEndTime;
LPSTR pszSystem;
LPSTR pszStatus;
LPSTR pszTaskName;
LPSTR pszTaskDesc;
} QUERYLOG, *PQUERYLOG;
Hmmm, this rather looks like C.
pQueryLog = (PQUERYLOG)GlobalLock(hMem);
nCnt = (int)(GlobalSize(hMem)/sizeof(QUERYLOG));
nCnt--; // zero based array
// * we have the memory, now copy the task.
pQueryLog[nCnt].pszStartDate = _strdup(szStartDate);
pQueryLog[nCnt].pszStartTime = _strdup(szStartTime);
[...]
GlobalUnlock(hMem);
What's the point of using GlobalLock() and GlobalUnlock()?
// * To use the data I do this:
pQueryLog = GlobalLock(hQueryMem);
nRecCnt = (int)(GlobalSize(hQueryMem)/sizeof(QUERYLOG));
// * Insert records into ListView
hWndLV = GetDlgItem(hWndDlg, LV_HISTORY);
for(i=0; i <= nRecCnt; i++)
This looks to me like a buffer overflow, if there are no elements (i.e.
nRecCnt==0) you access one element still.
memset(&lvItem,0,sizeof(lvItem));
lvItem.mask = LVIF_TEXT;
lvItem.cchTextMax = 12;
lvItem.iItem = i;
lvItem.iSubItem = 0;
lvItem.pszText = pQueryLog[nCnt].pszStartDate;
SendMessage(hWndLV,LVM_INSERTITEM,0,(LPARAM)&lvItem);
Beware, the win32 API is TCHAR based while your code assumes CHARs! This
might be a problem in the future but unless you are stupidly applying
casts, the compiler will warn you.
The data displays the exact same way regardless of how I save or
reference the data... I get garbage!!
Divide and conquer! Use OutputDebugString() to output the strings so you can
guarantee that they are right. Use the LVM_INSERTITEM message with fixed
strings to make sure those work.
Uli
===============================================
half frog half prince at oohay dot com
===============================================