I'm using this Delphi 7 code to detect if Internet Explorer is running:
function IERunning: Boolean;
begin
Result := FindWindow('IEFrame', NIL) > 0;
end;
This works on 99% of the systems with IE 8,9 and 10.
But there are some systems (unfortunately none of mine, but I have two beta testers which have such systems, both Win7 x64 SP1) where FindWindow() returns 0 for IEFrame, even if IE is in memory.
So I've coded an alternate method to find the window:
function IERunningEx: Boolean;
var WinHandle : HWND;
Name: array[0..255] of Char;
begin
Result := False; // assume no IE window is present
WinHandle := GetTopWindow(GetDesktopWindow);
while WinHandle <> 0 do // go thru the window list
begin
GetClassName(WinHandle, #Name[0], 255);
if (CompareText(string(Name), 'IEFrame') = 0) then
begin // IEFrame found
Result := True;
Exit;
end;
WinHandle := GetNextWindow(WinHandle, GW_HWNDNEXT);
end;
end;
The alternate method works on 100% of all systems.
My question - why is FindWindow() not reliable on some of the systems?
I'm guessing that FindWindow is declared to return a WinHandle, which is a THandle, which is an Integer, which is signed. (At least, I think this was the case many years ago when I programmed in Delphi.)
If IE has a window handle with the top bit set then it will be negative so your test will return False:
Result := FindWindow('IEFrame', NIL) > 0;
Window handles don't usually have the top bit set, but I don't know that it's impossible.
According to Delphi Help, FindWindow(ClassName,WindowName) does not search child windows. This could be the reason for the 1% failures. Maybe in those two beta tester's systems the IEFrame window had WS_CHILD style set.
This would explain why the GetTopWindow/GetNextWindow loop works. GetTopWindow(hWndParent) retrieves the child window at the top of the Z order, and GetNextWindow(hWnd,Direction) retrieves the next child window in the Z order.
This could be tested by calling FindWindowEx(hWndParent,hWndChild,ClassName,WindowName),
to see if it works where FindWindow() fails.
Related
In old Delphi applications which use the old and deprecated but still used BDE database engine with Paradox database files residing on a Windows 10 computer that's updated to the 1803 "Spring Creators Update" version, but the client computers using any older version of Windows like Windows 10 1709 or Windows 7, opening a Paradox table sometimes fails with a "No more files" error, idapi32.dll error code DBIERR_OSENMFILE. This raises a EDBEngineError exception in DBTables.pas / TTable.GetHandle(), which is called by TTable.CreateHandle, called by TBDEDataSet.OpenCursor().
The error seems to be caused by some file-sharing related changes in the Windows 10 1803 update. Removing the 1803 update from the file-sharing Windows 10 computer, or updating all the client computers to Windows 10 + 1803 seems to make the error go away.
People have speculated that the changes have something to do with the SMB protocol, maybe Windows Defender and/or other security related issues. Here's a Google Plus discussion
https://plus.google.com/106831056534874810288/posts/F4nsoTz2pDi
How could the "No more files" error be worked around by some reasonably easily doable changes in the Delphi application, while allowing the file-sharing client and server computers to keep using heterogeneous Windows versions?
Please try to refrain from answering or commenting self-evident things like "the sky is blue" or "BDE is old and deprecated". Keeping BDE is a decision that cannot be changed, certainly not as a "bug fix".
As an emergency fix, we have resorted to simply re-trying DbiOpenTable, when it returns the DBIERR_OSENMFILE error code. I posted an answer with source code to the idapi32.dll hack. So far it seems that if the first DbiOpenTable says "No more files", the second try succeeds, and the application works without noticing anything.
WARNING: what follows is a hack. A kludge. Band-aid, glue, duct tape and chewing gum. BDE is old. You are completely on your own if you use BDE and/or if you try this hack. I accept no responsibility over its use. If it works for you, good for you. If it ruins your business, bad for you.
Since the Paradox tables still mostly worked and the error seemed to be slightly randomly triggered, and since someone suspected Windows Defender having something to do with it, I thought maybe it just needs some kicking around. If DbiOpenTable() suddenly starts sometimes failing over a certain combination of SMB client/server versions, because "No more files" ... then why not just try the file operation again. I put an "if it returns a DBIERR_OSENMFILE error, then Sleep() and try again" logic around the DbiOpenTable function, and guess what - it seemed to work.
Hacking around the BDE's "features" is familiar to anyone who has to maintain BDE based applications. So I made a patching hook around idapi32.dll's DbiOpenTable function, starting from an old routine written by Reinaldo Yañez originally to fix the "insufficient disk space" error with BDE when the free disk space is at a 4 GB boundary. See https://cc.embarcadero.com/Item/21475
To use this, add Fix1803 in a uses clause, and call PatchBDE somewhere before starting to open Paradox tables. Maybe call UnPatchBDE when you're done, though I don't think that's necessary.
But remember, you're on your own, and this is highly experimental code.
unit Fix1803;
// * KLUDGE WARNING *
// Patch (hack) idapi32.dll DbiOpenTable() to try harder, to work with Windows 10 1803 "Spring Creators Update".
//
// The patching routine is an extension of code originally written by Reinaldo Yañez.
// see https://cc.embarcadero.com/Item/21475
//
// Some original Spanish comments are left in place.
interface
procedure PatchBDE;
procedure UnPatchBDE;
implementation
uses
Windows, Db, DbTables, BDE, SysUtils;
// ------------------------------------------- DbiOpenTable hook
var DbiOpenTable_address_plus_9 : Pointer;
function Actual_DbiOpenTable_CallStub(hDb: hDBIDb; pszTableName: PChar; pszDriverType: PChar; pszIndexName: PChar; pszIndexTagName: PChar; iIndexId: Word; eOpenMode: DBIOpenMode; eShareMode: DBIShareMode; exltMode: XLTMode; bUniDirectional: Bool; pOptParams: Pointer; var hCursor: hDBICur): DBIResult stdcall; assembler;
asm
// these two instructions are implicitly contained in the start of the function
// push ebp
// mov ebp, esp
add esp, $fffffee8
jmp dword ptr [DbiOpenTable_address_plus_9]
end;
function LogHook_DbiOpenTable (hDb: hDBIDb; pszTableName: PChar; pszDriverType: PChar; pszIndexName: PChar; pszIndexTagName: PChar; iIndexId: Word; eOpenMode: DBIOpenMode; eShareMode: DBIShareMode; exltMode: XLTMode; bUniDirectional: Bool; pOptParams: Pointer; var hCursor: hDBICur): DBIResult stdcall;
var
i : Integer;
begin
Result := Actual_DbiOpenTable_CallStub(hDb, pszTableName, pszDriverType, pszIndexName, pszIndexTagName, iIndexId, eOpenMode, eShareMode, exltMode, bUniDirectional, pOptParams, hCursor);
// if we got the "No more files" error, try again... and again.
i := 1;
while (Result = DBIERR_OSENMFILE) and (i < 10) do
begin
Windows.Sleep(i);
Result := Actual_DbiOpenTable_CallStub(hDb, pszTableName, pszDriverType, pszIndexName, pszIndexTagName, iIndexId, eOpenMode, eShareMode, exltMode, bUniDirectional, pOptParams, hCursor);
Inc(i);
end;
end;
// ------------------------------------------- Patching routines
const // The size of the jump instruction written over the start of the original routine is 5 bytes
NUM_BYTES_OVERWRITTEN_BY_THE_PATCH = 5;
type
TRYPatch = record
OrgAddr: Pointer;
OrgBytes: array[0..NUM_BYTES_OVERWRITTEN_BY_THE_PATCH-1] of Byte;
end;
procedure TRYPatch_Clear(var ARYPatch : TRYPatch);
begin
FillChar(ARYPatch, SizeOf(TRYPatch), 0);
end;
function RedirectFunction(OldPtr, NewPtr, CallOrigStub : Pointer; var OriginalRoutineAddressPlusN: Pointer; NumBytesInCompleteInstructionsOverwritten : Integer): TRYPatch;
type
PPtr=^pointer;
PPPtr=^PPtr;
TByteArray=array[0..maxint-1] of byte;
PByteArray=^TByteArray;
function SameBytes(Ptr1, Ptr2 : Pointer; NumBytes : Integer) : Boolean;
var
i : Integer;
begin
Result := true;
i := 0;
while (Result) and (i < NumBytes) do
begin
Result := Result and ((PByteArray(Ptr1)^[i] = PByteArray(Ptr2)^[i]));
Inc(i);
end;
end;
var
PatchingAddress : Pointer;
OldProtect,
Protect : DWORD;
p: PByteArray;
i : Integer;
begin
PatchingAddress := OldPtr;
if PWord(PatchingAddress)^ = $25FF then
begin {Es un JMP DWORD PTR [XXXXXXX](=> Esta utilizando Packages)}
p := PatchingAddress;
PatchingAddress := (PPPtr(#p[2])^)^; // PatchingAddress now points to the start of the actual original routine
end;
// Safety check (as if this thing was "safe"). The given replacement routine must start with the same bytes as the replaced routine.
// Otherwise something is wrong, maybe a different version of idapi32.dll or something.
if (CallOrigStub <> nil) and not SameBytes(PatchingAddress, CallOrigStub, NumBytesInCompleteInstructionsOverwritten) then
raise Exception.Create('Will not redirect function, original call stub doesn''t match.');
// Change memory access protection settings, so we can change the contents
VirtualProtect(PatchingAddress, NUM_BYTES_OVERWRITTEN_BY_THE_PATCH, PAGE_READWRITE, #OldProtect);
// Save the old contents of the first N bytes of the routine we're hooking
Result.OrgAddr := PatchingAddress; // Save the address of the code we're patching (which might not be the same as the original OldPtr given as parameter)
for i := 0 to NUM_BYTES_OVERWRITTEN_BY_THE_PATCH-1 do
result.OrgBytes[i] := PByte(Integer(PatchingAddress) + i)^;
// Replace the first bytes of the original function with a relative jump to the new replacement hook function
// First write the instruction opcode, $E9 : JMP rel32
PByte(PatchingAddress)^:= $E9;
// Then write the instruction's operand: the relative address of the new function
PInteger(Integer(PatchingAddress)+1)^ := Integer(NewPtr) - Integer(PatchingAddress) - 5;
// Address to jump to, for the replacement routine's jump instruction
OriginalRoutineAddressPlusN := Pointer(Integer(PatchingAddress) + NumBytesInCompleteInstructionsOverwritten);
// Restore the access protection settings
VirtualProtect(PatchingAddress, NUM_BYTES_OVERWRITTEN_BY_THE_PATCH, OldProtect, #Protect);
FlushInstructionCache(GetCurrentProcess, PatchingAddress, NUM_BYTES_OVERWRITTEN_BY_THE_PATCH);
end;
procedure RestorePatch(RestorePatch: TRYPatch);
var
OldProtect,
Protect : DWORD;
OldPtr: Pointer;
i : Integer;
begin
OldPtr := RestorePatch.OrgAddr;
VirtualProtect(OldPtr, NUM_BYTES_OVERWRITTEN_BY_THE_PATCH, PAGE_READWRITE, #OldProtect);
for i := 0 to NUM_BYTES_OVERWRITTEN_BY_THE_PATCH-1 do
PByte(Integer(OldPtr) + i)^ := RestorePatch.OrgBytes[i];
VirtualProtect(OldPtr, NUM_BYTES_OVERWRITTEN_BY_THE_PATCH, OldProtect, #Protect);
FlushInstructionCache(GetCurrentProcess, OldPtr, NUM_BYTES_OVERWRITTEN_BY_THE_PATCH);
end;
var
idapi32_handle: HMODULE;
Patch_DbiOpenTable : TRYPatch;
procedure PatchBDE;
begin
if idapi32_handle <> 0 then Exit; // already_patched
idapi32_handle := LoadLibrary('idapi32');
if idapi32_handle <> 0 then
begin
Patch_DbiOpenTable := RedirectFunction(GetProcAddress(idapi32_handle, 'DbiOpenTable'), #LogHook_DbiOpenTable, #Actual_DbiOpenTable_CallStub, DbiOpenTable_address_plus_9, 9);
end;
end;
procedure UnPatchBDE;
begin
if idapi32_handle <> 0 then
begin
{Leave everything as before, just in case...}
if Patch_DbiOpenTable.OrgAddr <> nil then
RestorePatch(Patch_DbiOpenTable);
FreeLibrary(idapi32_handle);
idapi32_handle := 0;
end;
end;
initialization
idapi32_handle := 0;
TRYPatch_Clear(Patch_DbiOpenTable);
end.
VMWare, Virtual Box, etc to virtualize an Windows 7. If, as you say, W7 work flawlessly that would solve the problem.
Our programming dept just spent about a non-mythical man-month tracking down what we think is a bug in a 3rd party component, here's their copyrighted source code:
function TGDIPPicture.GetImageSizes: boolean;
var
multi: TGPImage;
pstm: IStream;
hGlobal: THandle;
pcbWrite: Longint;
begin
result := false;
if Empty then
Exit;
if FDataStream.Size = 0 then
Exit;
hGlobal := GlobalAlloc(GMEM_MOVEABLE, FDataStream.Size);
if (hGlobal = 0) then
raise Exception.Create('Could not allocate memory for image');
try
pstm := nil;
// Create IStream* from global memory
CreateStreamOnHGlobal(hGlobal, TRUE, pstm);
pstm.Write(FDataStream.Memory, FDataStream.Size,#pcbWrite);
multi := TGPImage.Create(pstm);
FWidth := multi.GetWidth;
FHeight := multi.GetHeight;
Result := true;
multi.Free;
finally
GlobalFree(hGlobal);
end;
end;
We found the problem was with TMS's AdvOfficeTabSet. If we added tabs, then it crashed, if we didn't add tabs then it didn't crash. (the crash was one of those un-debuggable app hangs that hits you 10 steps after the real problem).
Following Raymond Chen's advice I replaced GMEM_MOVEABLE with GPTR and it appears to have fixed the problem.
I'm wondering if anyone can tell me if the above code had any legitimate reason for using GMEM_MOVEABLE. AFAIK it's only for the clipboard and it should always be used with GlobalAlloc.
while I was typing this another programmer got an error in the GlobalFree function using my code. So, apparently this doesn't work either. Could really use some help here!
*CreateStreamOnHGlobal is a Windows API function. (which apparently prefers GMEM_MOVEABLE)
*TGPImage is part of TMS's implementation of the GDI+ library.
Jonathan has identified the obvious problem, that being the double free of the HGLOBAL. But as you have found, the use is GMEM_MOVEABLE is correct.
Frankly, the code seems needlessly complex. I suggest you use the built in stream adapter and avoid any GlobalAlloc. To get an IStream you just need to do this:
pstm := TStreamAdapter.Create(FDataStream);
That's it.
I've got a Delphi 7.0 application that throws a memory access exception / message box every time it writeln's an empty string from the string list associated with a combo box:
csvstrlst := combobox1.Items;
csvstrlst.clear;
csvstrlst.add(''); //problem
csvstrlst.add('a'); //no problem
csvstrlst.add(''); //problem
csvstrlst.add('b'); //no problem
//throws memory access messages (I think the writeln writes a line though)
for n := 1 to csvstrlst.Count do begin
writeln(out_file,csvstrlst.strings[n-1])
end;
//throws memory access messages (writeln does write a comma text string though)
writeln(out_file,csvstrlst.commatext);
Running on Windows 7 or XP. As application or in D7 IDE. Combobox with empty string items also causes the same error if the parent of the form it is on is changed.
Has anyone else ever seen or heard of this problem? Any other information available at all?
This is a known and solved bug described in QC:
TCombobox gives AV when selecting empty item from dropdown
Although this is a bug, you should not reuse parts from controls to perform some data tasks as described in your question.
You will not save anything doing so, but getting most the time unwanted sideeffects (controls get repainted and/or fire events)
If you want to have a TStringList then create an instance.
csvstrlst := TStringList.Create;
try
// csvstrlst.Clear;
csvstrlst.Add( '' );
csvstrlst.Add( 'a' );
csvstrlst.Add( '' );
csvstrlst.Add( 'b' );
for n := 0 to csvstrlst.Count - 1 do
begin
WriteLn( out_file, csvstrlst[n] )
end;
WriteLn( out_file, csvstrlst.CommaText );
finally
csvstrlst.Free;
end;
As Sir Rufo has discovered the issue is a VCL bug introduced in Delphi 7 as described in QC#2246. According to that report the bug is resolved in a build with major version number 7 so you may be able to fix the problem by applying the latest Delphi 7 updates.
If not then you can fix the problem from the outside. I don't actually have a Delphi 7 installation to test this on, but I believe that this interposer class will work.
type
TFixedComboBoxStrings = class(TComboBoxStrings)
protected
function Get(Index: Integer): string; override;
end;
TComboBox = class(StdCtrls.TComboBox)
protected
function GetItemsClass: TCustomComboBoxStringsClass; override;
end;
function TFixedComboBoxStrings.Get(Index: Integer): string;
var
Len: Integer;
begin
Len := SendMessage(ComboBox.Handle, CB_GETLBTEXTLEN, Index, 0);
if (Len <> CB_ERR) and (Len > 0) then
begin
SetLength(Result, Len);
SendMessage(ComboBox.Handle, CB_GETLBTEXT, Index, Longint(PChar(Result)));
end
else
SetLength(Result, 0);
end;
function TComboBox.GetItemsClass: TCustomComboBoxStringsClass;
begin
Result := TFixedComboBoxStrings;
end;
The bug that was introduced in Delphi 7 is simply that the if statement reads:
if Len <> CB_ERR then
So, when Len is zero, that is when the item is the empty string, the True branch of the if is chosen. Then, the SendMessage becomes:
SendMessage(ComboBox.Handle, CB_GETLBTEXT, Index, Longint(PChar('')));
Now, PChar('') has special treatment and evaluates to a pointer to read only memory containing a zero character. And so when the combo box window procedure attempts to write to that memory, an access violation occurs because the memory is read only.
I'm writing an app I'd like to be backwardly compatible to some extent on XP, or at the very least windows vista.
EDIT FOR CLARITY: I need to be able to do what the first code snippet below does, but in XP. "Does anybody know the best approach to take under XP, given the functions aren't available in USER32.DLL.?"
My initial prototype code on windows 7 just called CreateProcess to start up displayswitch.exe, which is deployed with windows 7.
if you are not familiar with it, it's a handy little utility that is what gets invoked when you press the windows key and the letter P. you can read more about it here.
while this was adequate, i subsequently needed to sense the current state (eg internal vs external or extend vs clone), so i have now coded up a winapi solution that works well on windows 7 (and i presume 8). it involves making calls to SetDisplayConfig and QueryDisplayConfig in User32.DLL
The pertinent section of it is here (minus the many, many structures i had to hand craft in pascal code from the original klingon).
function getTopology : DISPLAYCONFIG_TOPOLOGY_ID ;
var NumPathArrayElements,
NumModeInfoArrayElements : UINT32;
var PathArrayElements_Size,
ModeInfoArrayElements_Size : UINT32;
error : Longint;
paths : PDISPLAYCONFIG_PATH_INFO_array;
info : PDISPLAYCONFIG_MODE_INFO_array;
begin
NumModeInfoArrayElements := 0;
Result := DISPLAYCONFIG_TOPOLOGY_EXTERNAL;
inc(result);
error := GetDisplayConfigBufferSizes(QDC_DATABASE_CURRENT,NumPathArrayElements,NumModeInfoArrayElements);
case error of
ERROR_SUCCESS :
begin
PathArrayElements_Size := sizeof(DISPLAYCONFIG_PATH_INFO) * NumPathArrayElements ;
ModeInfoArrayElements_Size := sizeof(DISPLAYCONFIG_MODE_INFO) * NumModeInfoArrayElements;
GetMem(paths,PathArrayElements_Size);
try
GetMem(info,ModeInfoArrayElements_Size );
try
error := QueryDisplayConfig(QDC_DATABASE_CURRENT,NumPathArrayElements, paths,NumModeInfoArrayElements, info,result);
case error of
ERROR_SUCCESS :;
else
Result := DISPLAYCONFIG_TOPOLOGY_EXTERNAL;
inc(result);
end;
finally
FreeMem(info,ModeInfoArrayElements_Size );
end;
finally
FreeMem(paths,PathArrayElements_Size);
end;
end;
end;
end;
function setTopology ( top : DISPLAYCONFIG_TOPOLOGY_ID) : boolean;
var flags : dword;
begin
result := false;
flags := DecodeDISPLAYCONFIG_TOPOLOGY_ID_SDC(top);
if flags <> 0 then
begin
result := SetDisplayConfig(0,nil,0,nil,SDC_APPLY or flags) = ERROR_SUCCESS;
end;
end;
Since these functions don't exist in XP (as far as I know), I am looking for a stable way of achieving a similar thing in XP. whilst i am coding in Delphi, it's not necessary that the solution be presented as such. i am quite happy to just look at how it's done, or read a description of the appropriate steps, and implement it myself.
(removed full listing as it was confusing the issue as it did not appear like a question)
I have a few old apps I've written (in Delphi) which for various reasons use a system tray icon. Most are using AppControls TacTrayIcon or some other similar component.
Here's my question: How does one control the position of a tray icon? (i.e. where it is, say, relative to the system time -- 1st position/"slot", 2nd position/"slot", etc). I recall seeing a demo (C#, if memory serves) that allowed the user to "shift icon to the left" and "shift icon to the right", but don't recall how it was done.
I'd like to allow the user to select what position they want to icon to appear in, for Windows 2000 - Windows 7. (I understand Windows 7 handles system tray stuff a little differently, but haven't tested that out yet).
Thanks for any and all help.
There is no documented or supported way for programs to control the positions of their shell notification icons. There's not even anything guaranteeing they will appear at all, or if they do appear, that they will appear anywhere near the clock, such that your positioning instructions would make any sense.
(I used to use a program that hijacked some or all of the icons and optionally displayed them in its own window instead of in the area near the clock. It was TraySaver, by Mike Lin. The source is available if you wish to see how his hack worked.)
The icon's position is not under your control. My advice to you is to not try to make it your program's responsibility, especially if nobody has actually requested such functionality from your program in the first place. If people want to control your program's icon location, they probably want to control other programs' icon locations, in which case the problem is bigger than you anyway.
Accessing and modifying the shell notification area is hackish but possible. You first need to find the top level window:
var
Wnd: HWND;
begin
Wnd := FindWindow('Shell_TrayWnd', nil);
if IsWindow(Wnd) then
EnumChildWindows(Wnd, #FindTrayWnd, 0);
end;
then enumerate its children to find the tray notification area:
function FindTrayWnd(AWnd: HWND; AParam: LPARAM): BOOL; stdcall;
var
ClassName: string;
begin
SetLength(ClassName, 64);
SetLength(ClassName, GetClassName(AWnd, PChar(ClassName), 64));
Result := True;
if AnsiCompareText(ClassName, 'TrayNotifyWnd') = 0 then begin
EnumChildWindows(AWnd, #FindToolbar, 0);
Result := False;
end;
end;
then enumerate its children to find the standard Windows toolbar with the notification icons. Windows messages are used to get or set toolbar properties. Since the toolbar lives in another process you need to employ ReadProcessMemory() and WriteProcessMemory() for all messages that involve a buffer of some sort (like getting the button text or button info):
function FindToolbar(AWnd: HWND; AParam: LPARAM): BOOL; stdcall;
const
VMFLAGS = PROCESS_VM_OPERATION or PROCESS_VM_READ or PROCESS_VM_WRITE;
var
ClassName: string;
i, ButtonCount: integer;
ProcessId, BytesRead: Cardinal;
ProcessHandle: THandle;
ExplorerButtonInfo: PTBButton;
ButtonInfo: array of TTBButton;
begin
SetLength(ClassName, 64);
SetLength(ClassName, GetClassName(AWnd, PChar(ClassName), 64));
if AnsiCompareText(ClassName, 'ToolbarWindow32') = 0 then begin
GetWindowThreadProcessId(AWnd, #ProcessId);
ProcessHandle := OpenProcess(VMFLAGS, FALSE, ProcessId);
ExplorerButtonInfo := VirtualAllocEx(ProcessHandle, nil, SizeOf(TTBButton),
MEM_RESERVE or MEM_COMMIT, PAGE_READWRITE);
if ExplorerButtonInfo <> nil then try
ButtonCount := SendMessage(AWnd, TB_BUTTONCOUNT, 0, 0);
SetLength(ButtonInfo, ButtonCount);
for i := 0 to ButtonCount - 1 do begin
SendMessage(AWnd, TB_GETBUTTON, i, LPARAM(ExplorerButtonInfo));
ReadProcessMemory(ProcessHandle, ExplorerButtonInfo, #ButtonInfo[i],
SizeOf(TTBButton), BytesRead);
end;
// manipulate the button info, use WriteProcessMemory() and SendMessage()
// to repopulate the toolbar
finally
VirtualFreeEx(ProcessId, ExplorerButtonInfo, SizeOf(TTBButton),
MEM_RELEASE);
end;
Result := False;
end else
Result := True;
end;
You should be able to identify the button of your notification icon via its name, then delete that button, then insert it at the desired position. All error handling omitted, but this should get you started.
Look at this article on CodeProject, hope it helps.
This solution have very limited compatibility, and not recommended at all.