Forum

How can I get a CExtDateTimeWnd control to display in the regional long date format?

I am using the CExtGridCellInplaceSlider in a CExtGridWnd. I monitor changes in cell value by overriding OnScrollPosSet. If I drag the slider or use Alt+left right arrow keys this function is called with pWndGrid valid which has been working fine. I recently noticed that if I set the value directly by entering the number OnScrollPosSet is called with pWndGrid set to NULL. I need pWndGrid to ultimately get the pointer to the document that needs to be altered. It appears that the only way I can do that in the case of direct entry is by overriding OnInplaceControlTextInputComplete. Is there some other central location that I might catch where the scroll position is being set that includes a valid pWndGrid so that I do not have to override two functions?

Please look at the following method declaration in the CExtGridCellInplaceSlider class:

virtual ULONG ScrollPosSet( ULONG nScrollPos );
virtual ULONG OnScrollPosSet(
      ULONG nScrollPos,
      CExtGridWnd * pWndGrid = NULL,
      LONG nColNo = 0L,
      LONG nRowNo = 0L,
      INT nColType = 0,
      INT nRowType = 0
      );

Here is the source code of the ScrollPosSet() method which simply calls OnScrollPosSet():

ULONG CExtGridCellInplaceSlider::ScrollPosSet( ULONG nScrollPos )
{
      ASSERT_VALID( this );
      return OnScrollPosSet( nScrollPos );
}

That means you can call directly the OnScrollPosSet() method and specify all the needed parameters.

Thank you for the response.

I looked further and would like to suggest that you change the source to CExtGridCellInplaceSlider::OnInplaceControlTextInputComplete as follows:

// Remove
//            ScrollPosSet( nVal );

// Put in it’s place
     OnScrollPosSet(
         nVal,
         &wndGrid,
         nColNo,
         nRowNo,
         nColType,
         nRowType );

Currently OnScrollPosSet is hit if an individual uses direct entry, drags the slider or uses Alt+arrow keys however in the case of direct entry not all of the parameters are filled in if direct entry is used. Why use ScrollPosSet when it is possible to directly call OnScrollPosSet in OnInplaceControlTextInputComplete. This way a user of this cell type can catch a changed value by overriding one method rather than two and always count on having the parameters filled in. PS - ScrollPosSet is NOT called when a user drags the slider or uses the Alt+arrow keys anyway and even so overriding it is likely to have little value if cells don’t know where they are or what grid they are in.

Just a suggestion.

Are there any example of how to add a toolbar to a dialog? I’ve tried, but I only get one button in the middle of the toolbar.

OK, thanks. I have the toolbar in the dialog now. Now I have a problem where the ON_UPDATE_COMMAND_UI messages are not beeing called until a button is clicked. After setting false for pCmdUI->Enable, the button can never reenabled.

The command updating mechanism for toolbars and menus used in the dialog window should work exactly like in the frame window and typically you should not do anything special for that. Toolbar buttons are updated by the toolbar’s owner window. By default, the toolbar’s owner window is its parent dialog or frame window. The toolbar’s owner window is detected automatically, but in some cases like non-modal dialogs, you may need to specify the toolbar’s owner window explicitly by invoking its CWnd::SetOwner() method. Please let us know more details about the dialog window in your project or send us a test project so we can help you find out what is wrong.

Hi,

I’m a new Prof-UIS user and I’m struggling to understand how I’m meant to deallocate memory with Prof-UIS derived classes, in particular with a CExtPropertyItem derived class.

In all past projects I have worked on the golden rule has been that if I instantiate an object on the heap with operator ’new’ then it is my responsibility to release memory for that object via a destructor call. If an object is instantiated in a third-party library then such an object is normally obtained via a Create() function and released via a Destroy() or Delete() function.

I can find no instructions in the Prof-UIS documentation which tell me which objects are my responsibility to delete, and which objects will be deleted by Prof-UIS. I therefore assumed that I would have to delete any objects I create on the heap via ’new’. However when I study the source of your samples for CExtPropertyGridCtrl I note that you create a CExtPropertyStore object and explicitly make comments in the code that this object should be deleted as its memory was allocated dynamically. You then go on to create many objects derived from the CExtPropertyValue class, and you instantiate these via operator ’new’ but you *don’t* call delete on these. Why not? Nowhere in the docs does it say that the CExtPropertyStore parent of these objects will handle the memory deallocation. How are Prof-UIS users meant to know which objects they should delete?

Also I don’t understand the usage of the CExtPropertyItem.Delete() method. You say that this should be called rather than using the ’delete’ operator, but what if I create a CExtPropertyValue-derived class which makes its own heap allocations? How do I destroy this object? Will CExtPropertyItem.Delete() call my own destructor, or should I first call the Destroy() method to release base class memory, and then destroy my own objects separately?

I apologise if this seems so trivial to the authors of the Prof-UIS library, and I realise I am probably misunderstanding this issue, but due to the dynamic nature of the app I am working on it is extremely important to me that I know exactly how to release memory from the numerous objects I am creating.

Many thanks for your time.

Andrew

Dear Support,

I’ve been trying to use your CExtGridWnd::GridColumnDefaultValueBind method. When specifying a column index >= zero, everything works fine. Your Help file says if I set first parameter (column index) as a negative value, the method will bind the given cell to all columns. I tried this version (would save me a loop), but any subsequent GridCellGet call with NULL as parameter for CRuntimeClass failed.
Stepping through your code for CExtGridWnd::GridColumnDefaultValueBind, I can’t see where it handles negative column index value differently from non-negative values. Could you please check this issue? Or would you like me to make a small sample application for demonstrating this problem?

Thank you!

We have replaced a CMenuBar with a CExtMenuControlBar.

We used to performe some customer specific operation (updating the menu depending on the other products installed) in the OnInitMenu of our application.

How can we do it again ?

Regards

There is an article describes how to build popup menus dynamically at run time.

OK let’s make things clear. It is not a pop-up menu.

We have replaced the MFC CMenuBar with a CExtMenuControlBar as described in the help delivered with ProfUI.

Previoulsy we used the OnInitMenu() to perform various action such as if the use has the Shift key down we displayed some special menu.

This no longer works since the OnInitMenu is never reached with the MenuBar.

The LRESULT CMainFrame::OnExtMenuPrepare(WPARAM wParam, LPARAM lParam) as described in the article just doesn’t work (catch it for the popup but not the mainframemenu)

How can we fix this?

Hi,

I’m looking for the proper way to free/delete all the memory used by EnableCustomization(). When I call the following code, I get a bunch of memory leaks when shutting down.

if(!CCustomizeDlg::EnableCustomization(this,__ECSF_DEFAULT)) {
ASSERT( FALSE );
return -1;
}

Is there a shutdown function I should be calling in DestroyWindow()?

Thanks,
Mike

Problem solved.

Added AfxOleInit(); to the CApp::InitInstance() no more leaks. Not sure why this is the case, but it works great and the 1000’s of leaks are plugged.

Thanks for the help. Prof-UI is a great tool.

Mike

The OLE initialization is essential because the customization subsystem uses OLE drag-and-drop APIs for menu items and toolbar buttons. But this should have to do with memory leaks. We guess the source of these leaks was specific to your project only.

There are no API that would allows you deinitialize the customization subsystem directly. Everything should be closed and deallocated automatically. If you have memory leaks of MFC/Prof-UIS objects, there must be some problem in the application shutdown. For instance, it may be closed by the ExitThread() or PostQuitMessage() APIs so the destructors won’t be called. Please check if the DestroyWindow() virtual method of the main frame class and the ExitInstance() virtual method of application class are invoked at shutdown. You can also configure the debugger to break on any exceptions before shutdown and clarify what happens during shutdown. Any unhandled exceptions may also crash your application even without any popup message boxes.

Dear Tech Support,

Is there a way to redraw the cell? Currently the text is updated only when the mouse is dragged over the cell.

The CExtGridCellString is inside a CExtGridWnd

Thanks

Thank you, your solution worked perfectly.

My application has several CExtCustomControlBars. If I pin one of them open and right click on the title bar (of the Controlbar) code gets stuck in an endless loop here:

CExtControlBar.cpp

void CExtControlBar::OnRButtonUp(UINT nFlags, CPoint point)
...
...
...
CWnd * pWndChild = GetWindow( GW_CHILD );
    if( pWndChild != NULL )
    {
        HWND hWndOwn = m_hWnd;
        if( stat_QueryFocusChangingEnabled( this, pWndChild->m_hWnd ) )
        {
            pWndChild->SetFocus();
            CExtPopupMenuWnd::PassMsgLoop( true ); <***** NEVER RETURNS FROM HERE
            if( ! ::IsWindow( hWndOwn ) )
            {
                delete pPopup;
                return;
            }
            m_bWindowActive = true;
            SendMessage( WM_NCPAINT );
        }
    }

The default context Prof-UIS menu is not overridden anywhere. It displays if click on un-pinned ControlBar title.
I do not see this problem in any of your samples.

Any idea what I might have done wrong?

Thanks

Hi

In our application we have to dynamically associate menu id with command (contextual menu).

The issue we face is that once a CmdID is associated with a function, it will remain.

Is there a way to clean this message map ?

Thanks

PageVisibleSet method don’t hide the hwnd attached with the page when PageVisibleSet is invoked to hide the item(it is clear when you try to hide the selected tab page). I already sent a mail with a demo project....
Also, I have another problem - seems that a CTreeCtrl embedded in a CExtResizableDialog attached in CExtTabPageContainerWnd does not show the scroll bar..... I will send you a demo project...

I sent the mail - thanks for help

Please find the following member of the CDlgDemoTreeCtrl class in your project:

      CExtWRB<CExtWFF< CTreeCtrl > > m_treeCtrl;

and replace it with

      CExtWFF < CTreeCtrl > m_treeCtrl;

You will also need to replace

CExtWRB<CExtWFF< CTreeCtrl > >& m_wndTree = m_treeCtrl;

code with the CTreeCtrl & m_wndTree = m_treeCtrl; code in the CDlgDemoTreeCtrl::OnInitDialog() method. The scroll bars will appear in the tree control after these modifications.

Please do not use the CExtWRB template class with the controls which have scroll bars. The CExtWRB template class re-computes and re-paints window non-client area making it looking like a thin 1 pixel border. But scroll bars of the tree control are not scroll bar windows. Thay are rather parts of the window non-client area and they are not based on HWND handles. So, the CExtWRB template class simply kills scroll bars.

The CExtWRB template class is designed for non-scrollable windows like CExtResizableDialog. When using CExtWRB with windows inside resizable control bars, the thin border looks like it is around many windows inside dockable panes in Visual Studio .NET and Visual Studio 2005. But the tree control already has a nice thin border.

I already sent the mail with treectrl page on Wednesday, August 15, 2007, subject ’Fw: CExtTabPageContainerWnd’. If you don’t have this mail tell me and I will send again....

No, we did not receive it. Could you send it again? Please also make sure that the zipped project is small enough, i.e. it does not contain redundant files like with .ncb extension.

Hi,

I want to set the icon display in right side of the button. So I search the help file.
I see a method "SetAlign" now is defined in your help file (ver 2.80), but I can’t use this function because this function is not defined in "ExttButton.h". And I found another same function "SetIconAlignment", but I don’t know how to use this function. Because I don’t know how to give the paramenter. Of cause, your help file without define this function.
Would you help me. Thanks

Thanks for your quick reply. I see the example too.
But I suggestion your help file need update at the next time release. It may very helpful for end-user. Otherwise we need search the sample code for our usage, I think it is not make sense. ˆˆ

I feel I need to let you know that I’m not part of the ProfUIS Support team. And even though I agree that Help file should be updated, I think it’s very important to let the users have the new feature as soon as possible (especially that this feature was requested).

Hello,

I tried using CExtResizablePropertySheet with CExtResizablePropertyPage. However, the title and subtitle are not appearing. Please help.

Regards

Can you support me on the issue? I have 2 user license and 90 days supporting.

Here are some codes from my project:
CCalibratorResizablePropertySheet wizard( _T("Calibrator"));
wizard.SetSliceImageData(pImageData);
// add pages
CCalibrateRawImagePropertyPage page1(&wizard);
wizard.AddPage( &page1 );
CCalibrateCheckDICOMPropertyPage page2(&wizard);
wizard.AddPage( &page2 );
CCalibrateCropImagePropertyPage page3(&wizard);
wizard.AddPage( &page3 );
CCalibrateContourPropertyPage page4(&wizard);
wizard.AddPage( &page4 );
CCalibrateResultPropertyPage page5(&wizard);
wizard.AddPage( &page5 );

// comment this line and next lines to set simple tabbed mode (non-wizard)
wizard.SetWizardMode();

// these lines are fully available with MFC 7.0 or later only
     // comment them to set old classic wizard mode
     // leave them intact to use wizard 97 mode
     wizard.m_psh.dwSize = sizeof(wizard.m_psh);
     wizard.m_psh.dwFlags |= (PSH_WIZARD97 |PSH_WATERMARK |PSH_WIZARDHASFINISH);
     // comment the following line if the tiled watermark bitmap is not needed
     wizard.m_psh.pszbmWatermark = MAKEINTRESOURCE(ID_CALIBRATE_WIZARD_WATERMARK);
     //wizard.m_psh.hInstance = ::AfxGetInstanceHandle();
     page1.m_psp.dwFlags |= PSP_DEFAULT|PSP_USEHEADERTITLE|PSP_USEHEADERSUBTITLE;
     page1.m_psp.pszHeaderTitle = _T("Title of page 1");
     page1.m_psp.pszHeaderSubTitle = _T("Subtitle of page 1");
     page2.m_psp.dwFlags |= PSP_DEFAULT|PSP_USEHEADERTITLE|PSP_USEHEADERSUBTITLE;
     page2.m_psp.pszHeaderTitle = _T("Title of page 2");
     page2.m_psp.pszHeaderSubTitle = _T("Subtitle of page 2");

I am crazy for the same issue. Can you tell me how to fix it? I use the propertysheet in MFC extensition DLL project.

The disappearing page title bug is a 50% bug. It depends on the property sheet initialization sequence. It’s absent in our ResizablePropertySheet sample application while it may present in your project. The CExtResizablePropertySheet class needs to anchor its child window because this is needed by its resizable features and it needs to subclass child controls with Prof-UIS classes and re-paint its surface for theme skinning purposes. These are some differences between the CExtResizablePropertySheet class in Prof-UIS and the CPropertySheet class in MFC. The CPropertySheet class does not create the inner content of the property sheet. This is performed by the PropertySheet() Win32 API. So, we are using a delayed initialization in the CExtResizablePropertySheet class. That is why it is very important to initialize it exactly like you we do it in the ResizablePropertySheet sample.

We still did not receive enough information about the problem. Would you send us your project so we can check it?

Hello,

I am having the following problems with Hyperlink and Property Sheets

1. Hyperlink has no hand cursor. Don’t know how to get it.

2. Property Sheet has no title or subtitle on following pages.

I have enclosed a project which demonstrates this (sent by email)

1. For compiling with just MFC (no ProfUIS), use the Build Configuration "Debug". Shows the title and the subtitle of the property page.

2. For compiling with ProfUIS, use the Build configuration "ProfUIS-Debug". Shows both problems listed above.

Thanks
Jay

You already replied your questions in other threads. Actually there is not enough information to find out what might be wrong. In fact, you simply say the problem exists and nothing else. Please send us your project demonstrating the problems or provide us with remote access to your desktop so we can help you.

Because there are no any problems with the hand cursor over hyperlink controls in our sample applications and there are no problems with titles in the ResizablePropertySheet sample. There must be something specific to your project.

I sent a sample project which illustrates the problem clearly. Since there is no way to include stuff in the forum replies, I have sent it by email. Did you get it? Looks like you have not got it yet.

If there is not enough information, that is because it is nothing complicated. Just create a dialog box project from scratch, replace the dialog box with a property sheet/page wizard and using your classes, the title and subtitle are absent.

Please let me know if you haven’t received the sample by email.

We received your e-mail, fixed the problems in your project and sent a modified version of your project back. The e-mail was sent 3:27PM EST on 8/14/2007. We suspect this email may have been filtered out or got into your Junk E-mail folder. So please check this. We can upload this email on our web site and give you the link or send to some other email address if you provide it to us.

There have been a number of discussions about having check boxes to enable/disabled a property value item (besides my original posts)

http://www.prof-uis.com/Forum_View.aspx?CID=40&M=55836&s=Property

http://www.prof-uis.com/Forum_View.aspx?CID=29&M=55995&s=property+check

What are your plans? Do you plan to continue to improve them, and make them a "public" part of the library?

Hello,
I’m currently using the powerful CExtReportGridWnd to display structured data and would like to walk the columns and rows in the order their have been left by the user in the grid, including collapsed rows.

Using ReportColumnGetStartPosition() and ReportColumnGetNext() enumerate the displayed columns in the order they were registered, not the user order.

Using the ItemGetRoot() and ItemGetNext() enumerate only the shown rows (not those which are collapsed when one or several fields are used as group criteria).

Could you please give hints to :
- walk columns in the order they are displayed in the grid ?
- walk rows in the order where they appear, even when not expanded ?

Thank you in advance,

Best Regards,

Christophe Guibert

This is how ItemGetNext is shown in Help file:

virtual HTREEITEM ItemGetNext(
   HTREEITEM hTreeItem,
   bool bSiblingOnly,
   bool bExpandedWalk,
   bool bWalkDeeper
) const;

Though I didnt try, but I guess if you use false as 3rd parameter, it should give you collapsed items as well. If it doesn’t, it might be a bug.

I thank you for your suggestion, but collapsed items are not traversed whatever the parameter I give, especially the third.

In Prof-UIS 2.80, there’s even a fourth boolean parameter (bIncludeHidden, not yet documented), and it seems to have no effect.
Is this a bug ?

The other point is to enumerate columns in the order they are set by the end-user. Any idea ?

Best Regards,

Christophe Guibert

The CExtTreeGridWnd::Item***() APIs are based on HTREEITEM handles and can be used for accessing any tree row. But if you want to access visible rows, you should use the CExtGridWnd class APIs (including its based classes) which are based on plain row indices. The CExtGridWnd::OnSiwGetVisibleRange() method returns a CRect object with the first/last displayed row indices in the CRect::top and CRect::bottom members. In order to convert the plain row indices into HTREEITEM handles and vice versa, you should use the CExtTreeGridWnd::ItemGetByVisibleRowIndex() and CExtTreeGridWnd::ItemGetVisibleIndexOf() methods. So, you can enumerate the entire scrollable range from zero up to CExtGridWnd::RowCountGet()-1 or visible row range returned by the CExtGridWnd::OnSiwGetVisibleRange() method.

Dear Technical Support,

Considering a tree like this one :
root:
-item 1
-child item 2
-child item 3
-item 4
-child item 5

When using ItemGetNext() from the root, it will walk the tree until reaching -child item3 with siblingOnly=false, then return null and not the item 4. Is this on purpose ?

Nevertheless, I’ve built a recursive function to walk the tree, using ItemGetNext and ItemGetParent , changing siblingOnly, and building a list of pointers to items. This satisfy my need.

Thank you for your support.

Best Regards

Christophe Guibert

I use this code to walk the columns in the order they appear on the screen.

	LONG lColNo = ColumnCountGet();
	for (LONG l = 0L; l < lColNo; l++)
	{
		CExtReportGridColumn *pRGC = (CExtReportGridColumn *)GridCellGetOuterAtTop(l, 0L);
		// do something here
	}

In my testing, it seems that ItemGetNext(hItem, false, false, true, false); gives just what you need. Can you reproduce the problem in a small sample application?

The example (for example "Load .." in "Compound Properties") have it, but could not find anyway of doing that. Please help.

The CExtHyperLinkButton class implements a hyperlink control. It is derived from CExtButton. That means the hyperlink is just a repainted button control. The CExtButton::SetBtnCursor() method allows you to assign a cursor to any kind of buttons in Prof-UIS (including CExtHyperLinkButton) which are derived from the CExtButton class. This method is invoked in the CExtHyperLinkButton class constructor and assigns a hand like cursor to it. So, each hyperlink control should display the hand cursor automatically.

Well Is there anything specific to be done to get this?

I have declared the hyperlink like this:

CExtHyperLinkButton m_purchaseNow;

In the dialog editor, I have it as a Button and in InitDialog(), I assign a URL to it.

Any other steps besides these?

Actually it is enough to get it working. You can compare your code with that used in the Prof-UIS Controls, where you can find an example of the hyperlink. Would you also tell us what is not working?

Hello,

As I mentioned, the link appears, but no cursor. How do I get this? I need this ASAP.

Regards

I tried looking at a few samples and I don’t see anything that is different. I have racked by brains about this. I looked in "Compound Properties" sample which has 4 hyperlinks each of which lead to actions within the program and also nicely change over to a hand cursor when I mouse-over. But same is not available to my program!

I am getting a memory fault in the following code when I try to use the property sheet: Works fine with just MFC property sheet. I replaced CPropertySheet with CExtResizablePropertySheet everywhere. No other changes. Please advise.


inline void __EXT_MFC_MEMCPY(
    void * dest,
    size_t sizeInBytes,
    const void * src,
    size_t count
    )
{
    #if _MFC_VER >= 0x0800
        memcpy_s( dest, sizeInBytes, src, count );
    #else
        memcpy( dest, src, count );
        sizeInBytes;
    #endif
}

You said you had replaced CPropertySheet with CExtResizablePropertySheet. Did you also replace CPropertyPage with CExtResizablePropertyPage?

Yes, I tried with both. Using only CExtResizablePropertySheet and using CExtResizablePropertySheet. with CExtResizablePropertyPage. However, I have not yet tried to replace *all* instances of CPropertyPage with CExtResizablePropertyPage, just the first. I wanted to get a look before I did all of them. Should I try replacing all of them? It is a quite bit of work, so I hesitated.

You should replace all instances otherwise it won’t work.

Dear Support,

I have some small buttons (CExtButton) which have only an icon on them, but no text. I want each of them to have focus rect when it is in focus. By default, focus rect is only around the text part of the button, but you have a CExtButton::SetFocusMode method, with which I can change this default behaviour and make the focus rect drawn around the client area of the button. But this still doesn’t allow me to have focus rect for a button without text. I’ve stepped through your code CExtPaintManagerXP::PaintPushButton and I see that the code which would draw the focus rect is inside an if-block with the following condition:

	if(		rectCaption.bottom > rectCaption.top
		&&	rectCaption.right > rectCaption.left
		)
	{ // if we have valid area for text & focus rect
		...
	}

So rectFocus is not checked, and even though I have valid rectFocus in that code, the invalid rectCaption keeps me from getting to the code which would draw the focus rect.

Please check this issue! Thank you!

Unfortunately you cannot do that.

The button is drawn in the CExtPaintManager::PaintPushButton() virtual method and in its overridden versions. Here is the code that draws the focus rectangle:

if(		_ppbd.m_bDrawFocusRect 
	&&	( ( !sTitle.IsEmpty() ) || _ppbd.m_nFocusMode == __EBFM_CLIENT_AREA )
	&&	rcFocus.bottom > rcFocus.top 
	&&	rcFocus.right > rcFocus.left
	)
{
	rcFocus.DeflateRect( 2, 2 );
	dc.DrawFocusRect( &rcFocus );
}

As for the forum, that you for this suggestions. We look into it, may be it will be not difficult to implement it.

I can also reproduce this problem in your ProfUIS_Controls sample.

Test #1:
Insert line
    m_btnPushButton.SetFocusMode(CExtPaintManagerXP::__EBFM_CLIENT_AREA);
in CPageButtons::OnInitDialog. Build, run sample, check CheckBox for "Show Focus Rect", press tab couple of times. Result: Focus rect is shown around client area of pushbutton.

Test #2:
Insert line
    m_btnPushButton.SetWindowText(_T(""));
right after the line with SetFocusMode we inserted in Test #1. Build, run sample, check CheckBox for "Show Focus Rect", press tab couple of times. Result: Focus rect is not shown anywhere inside pushbutton.

Please let me know if you encounter the same problem! Thank you.

The bug is fixed. Thank you again. You can download the updated source code from our ftp site.

Dear Support,

I haven’t compiled your new source code till now, so I couldn’t confirm the changes earlier. But now I see the fix is there, works nicely, thank you very much!

One more question concerning this issue: Our buttons are very-very small (18x18 pixel), the pictures in them cover almost the entire area of the button. Because of this, the focus rect is drawn over the picture. Is there a way I could change the distance between the button border and the focus rect? Only for these buttons, of course.

Chris

P.S.: Do you have any plans improving your forum in a way that the readers could sort the forum by date of last post? It’d be really great.

Hello,

we want to use CExtIconEditDlg to edit two color bitmaps.

We managed to get CExtIconEditDlg to work properly with a two color palette.

But we have problems to extract a two color HIBTMAP (bitmap with one bit per pixel) from the resulting m_icon member. What is the right way to do this?

First of all you should create a mono bitmap:

INT nWidth = . . .
INT nHeight = . . .
CBitmap bmpMono;
      VERIFY( bmpMono.CreateBitmap( nWidth, nHeight, 1, 1, NULL ) );

Then you should create a DC for it:

CDC dcMono;
      VERIFY( dcMono.CreateCompatibleDC( NULL ) );
CBitmap * pOldBitmapFromMonoDC = dcMono.SelectObject( & bmpMono );

You also have some bitmap from the image editor window. You should also create a DC for it and select this bitmap in its DC. Then you should draw using the BitBlt(),/code> API from the second DC into <code>dcMono. Finally you should restore the old selected bitmaps in both DCs and destroy DCs. As a result, you have a 2-BPP bitmap in bmpMono.

Hello,

many thanks for your response.

This is what we have done already:

CExtIconEditDlg dlg;
INT nWidth = . . .
INT nHeight = . . .

CBitmap bmpMono;
VERIFY( bmpMono.CreateBitmap( nWidth, nHeight, 1, 1, NULL ) );

dlg.m_icon.AssignFromHBITMAP((HBITMAP)bmpMono.GetSafeHandle(), (COLORREF)-1);

if (dlg.DoModal() == IDOK)
{
CDC dcMono;
VERIFY( dcMono.CreateCompatibleDC( NULL ) );
CBitmap * pOldBitmapFromMonoDC = dcMono.SelectObject( & bmpMono );

CBitmap oBmpDlg;
CBmpDlg.Attach(dlg.m_icon.m_bmpNormal.CreateBitmap(NULL, (COLORREF)-1));

CDC dcMemDlg;
dcMemDlg.CreateCompatibleDC(NULL);
CBitmap* poOldBitmapDlg = dcMemDlg.SelectObject(&oBmpDlg);

dcMono.BitBlt(0, 0, bmp.bmWidth, bmp.bmHeight, &dcMemDlg, 0, 0, SRCCOPY);

dcMemDlg.SelectObject(poOldBitmapDlg);
dcMono.SelectObject(pOldBitmapFromMonoDC);
}

but everything, what we finally receive in dcMono is a completely black bitmap, regardless, which image we created in the editor. What is wrong?

We are sure we provided you with the correct way of creating one bit per pixel bitmaps. Please update the CMainDlg::OnButtonEdit() method in the IconEditor sample.

void CMainDlg::OnButtonEdit() 
{
int nSel = m_wndList.GetCurSel();
      if( nSel < 0 )
      {
            ::AfxMessageBox(
                  _T("Please select icon in list"),
                  MB_OK|MB_ICONINFORMATION
                  );
            return;
      }
CExtCmdIcon * pIcon = m_wndList.GetIconPtr( nSel );
      ASSERT( pIcon != NULL && (! pIcon->IsEmpty()) );
CExtIconEditDlg dlg( this );
      dlg.m_icon.AssignFromOther( *pIcon );
      ASSERT( ! dlg.m_icon.IsEmpty() );
      dlg.m_strProfileSection = _T("ResizableDialogs");
      dlg.m_strProfileEntryWindow = _T("IconEditor-Window");
      dlg.m_strProfileEntryIconEditor = _T("IconEditor-EditorControl");
CSize _sizeIcon = pIcon->GetSize();
CString strColors;
INT nBitsPerPixel = pIcon->m_bmpNormal.GetBPP();
      if( nBitsPerPixel >= 24 )
            strColors = _T("16.777.216 (16M)");
      else if( nBitsPerPixel == 16 )
            strColors = _T("65.536 (64K)");
      else if( nBitsPerPixel == 15 )
            strColors = _T("32.768 (32K)");
      else
            strColors.Format( _T("%ld"), 1L << nBitsPerPixel );
      dlg.m_strAlternateCaption.Format(
            _T("Icon Editor - %dx%d, %s colors"),
            _sizeIcon.cx, _sizeIcon.cy,
            (LPCTSTR)strColors
            );
//    dlg.m_bEatTransparentColor = false;
      if( dlg.DoModal() != IDOK )
      {
///////           return;
      }
      ASSERT( ! dlg.m_icon.IsEmpty() );
      pIcon->AssignFromOther( dlg.m_icon );
      m_wndList.RecalcListLayout();

HBITMAP hBitmapColored = NULL;
COLORREF clrTransparent = COLORREF(-1L);
      VERIFY( dlg.m_icon.ExtractEditableBitmap( hBitmapColored, clrTransparent ) );
      ASSERT( hBitmapColored != NULL );
      VERIFY( CExtPaintManager::stat_SerializeBitmap( _T("C:\\1.bmp"), false, &hBitmapColored ) );

CExtBitmap bmpColored;
      bmpColored.FromBitmap( hBitmapColored );
      ::DeleteObject( hBitmapColored ); // we don’t need it anymore
CSize sizeBmp = bmpColored.GetSize();
int nCountOfBytesPerLine = sizeBmp.cx / 8 + ( ( ( sizeBmp.cx % 8 ) != 0 ) ? 1 : 0 );
int nCountOfBytesInBitmap = nCountOfBytesPerLine * sizeBmp.cy;
BYTE * pBinarySurface = new BYTE [ nCountOfBytesInBitmap ];
int nCounter, nX, nY;
      for( nCounter = 0; nCounter < nCountOfBytesInBitmap; nCounter ++ )
            pBinarySurface[ nCounter ] = BYTE(0xFF); // make each pixel white
double lfHue, lfSaturation, lfLuminance;
BYTE * pCurrSurfacePtr = pBinarySurface;
      for( nY = 0; nY < sizeBmp.cy; nY++ )
      {
            for( nX = 0; nX < sizeBmp.cx; nX++ )
            {
                  COLORREF clrColored = bmpColored.GetPixel( nX, nY );
                  if(         clrTransparent != COLORREF(-1L)
                        &&    clrTransparent == clrColored
                        )
                        continue;
                  CExtBitmap::stat_RGBtoHSL( clrColored, &lfHue, &lfSaturation, &lfLuminance );
                  if( lfLuminance > .5 )
                        continue;
                  int nByteIndex = nX / 8;
                  BYTE & _byte = pCurrSurfacePtr[ nByteIndex ];
                  int nBitIndex = nX % 8;
                  nBitIndex = 7 - nBitIndex;
                  BYTE nMaskValue = BYTE( 1 << nBitIndex ); 
                  _byte &= ~nMaskValue; // make this pixel black
            }
            pCurrSurfacePtr += nCountOfBytesPerLine;
      }
//HBITMAP hBitmapMono = CreateBitmap( sizeBmp.cx, sizeBmp.cy, 1, 1, pBinarySurface );
                  BITMAP bi;
                        bi.bmType = 0;
                        bi.bmWidth = sizeBmp.cx;
                        bi.bmHeight = sizeBmp.cy;
                        bi.bmWidthBytes = nCountOfBytesPerLine;
                        bi.bmPlanes = 1;
                        bi.bmBitsPixel = 1;
                        bi.bmBits = pBinarySurface;
                  HBITMAP hBitmapMono = CreateBitmapIndirect( &bi );
                        ASSERT( hBitmapMono != NULL );
                        delete [] pBinarySurface;

      VERIFY( CExtPaintManager::stat_SerializeBitmap( _T("C:\\2.bmp"), false, &hBitmapMono ) );
CExtBitmap bmpTmp;
bmpTmp.FromBitmap( hBitmapMono );
bmpTmp.Make32();
bmpTmp.SaveBMP_File( _T("c:\\3.bmp") );
      ::DeleteObject( hBitmapMono ); // we don’t need it anymore

}

Then please run it and open the C:\ folder with Windows Explorer and select the Thumbnails view for this folder. You can double click any list box item in the IconEditor sample and then click OK in the opened icon editor dialog, In the C:\ folder, you will see three bitmaps created in it. 1.bmp (the hBitmapColored and bmpColored variables in the code above) is created from the icon surface and uses some random color instead of transparent pixels. 2.bmp (the hBitmapMono) is absolutely an valid one pit per pixel bitmap file created from the previous bitmap but it is displayed absolutely incorrect for most of icons from the list box in the main dialog of the IconEditor sample. 3.bmp (the bmpTmp) is created from the second bitmap. This third bitmap is a 32 bit per pixel bitmap without alpha channel and contains black and white opaque pixels only. This bitmap is displayed correctly in the Windows Explorer.

We would like to ask you to provide more details about your task so we can find correct bitmap implementation for you.

Hello,

many thanks for your suggestion.

You wrote:
>>
This third bitmap is a 32 bit per pixel bitmap without alpha channel and contains black and white opaque pixels only.
<<
This is not exactly, what we need. We need a 1 bit per pixel bitmap as result.

Something like this:

HBITMAP hbmpSourceMono; // given bitmap with 1 bit per pixel
CExtIconEditDlg dlg( this );

dlg.m_icon.AssignFromHBITMAP(hbmpSourceMono, (COLORREF)-1);

if (dlg.DoModal() == IDOK)
{
HBITMAP hbmpDestMono = dlg.m_icon.Extract_HBITMAP_with_1_Bit_per_Pixel();
}

We are looking for that "Extract_HBITMAP_with_1_Bit_per_Pixel()" function, that results in a HBITMAP with 1 bit per pixel.

Or in other words: the complete procedure of editing bitmaps with CExtIconEditDlg should maintain the original color depth.

The source code in our previous message allows you to convert any source bitmap into a destination 1BPP bitmap (1 bit per pixel bitmap). If the source bitmap contains only RGB(0,0,0) and RGB(255,255,255) pixels, the destination bitmap will be exactly the same. We believe the generated 1BPP bitmap is correct. The only real problem is that it cannot be painted correctly by Windows GDI APIs for bitmap painting and even Windows Explorer does not display it correctly. Prof-UIS uses the same technique for generating 1 BPP bitmaps used for painting docked toolbar grippers in Office XP style. But our 1BPP bitmap is not painted using BitBlt() API. We use it for creating a pattern brush and fill the gripper area using it. The 1BPP pattern is painted correctly using this approach. So, we can offer you the following solution:

1) We will create a function from our code for converting any bitmap into a 1BPP bitmap. You could use it for converting after editing. So, you will keep only 1BPP bitmaps in memory.

2) Before painting and editing, you will convert 1BPP bitmaps into 32BPP bitmaps like it is demonstrated in the last lines of our code which generate the third bitmap.

Hello,

many thanks for your support.

>>
1) We will create a function from our code for converting any bitmap into a 1BPP bitmap. You could use it for converting after editing. So, you will keep only 1BPP bitmaps in memory.
<<
This would be great. Thank you very much in advance.

Woud it be also possible, to make the function more flexible and to allow converting any bitmap into a xBPP bitmap? In this way one could input any xBPP bitmap into CExtIconEditDlg and could output the resulting bitmap after DoModal() in the same color format.

Here is the function for creating a 1 BPP bitmap

HBITMAP ConvertBitmapToMono(
      HBITMAP hBitmapColored, // source bitmap of any format, must be not NULL
      bool bDeleteColoredBitmapOnSuccess = false, // delete hBitmapColored if the returned handle is not NULL
      double lfMinWhiteLuminance = 0.5 // luminance margin between 0.0 (black) and 1.0 (white)
      )
{
      if(         hBitmapColored == NULL
              ||  lfMinWhiteLuminance <= 0.0
              ||  lfMinWhiteLuminance >= 1.0
              )
        {
              ASSERT( FALSE );
              return NULL; // invalid parameter
        }
COLORREF clrTransparent = COLORREF(-1L);
CExtBitmap bmpColored;
      if( ! bmpColored.FromBitmap( hBitmapColored ) )
              return NULL;
CSize sizeBmp = bmpColored.GetSize();
int nCountOfBytesPerLine = sizeBmp.cx / 8 + ( ( ( sizeBmp.cx % 8 ) != 0 ) ? 1 : 0 );
int nCountOfBytesInBitmap = nCountOfBytesPerLine * sizeBmp.cy;
BYTE * pBinarySurface = NULL;
      try
      {
            pBinarySurface = new BYTE [ nCountOfBytesInBitmap ];
      }
      catch( CException * e )
      {
            ASSERT( FALSE );
            e->Delete();
            return NULL; // memory allocation error
      }
      catch( ... )
      {
            ASSERT( FALSE );
            return NULL; // memory allocation error
      }
      ASSERT( pBinarySurface != NULL );
int nCounter, nX, nY;
      for( nCounter = 0; nCounter < nCountOfBytesInBitmap; nCounter ++ )
            pBinarySurface[ nCounter ] = BYTE(0xFF); // make each pixel white
double lfHue, lfSaturation, lfLuminance;
BYTE * pCurrSurfacePtr = pBinarySurface;
      for( nY = 0; nY < sizeBmp.cy; nY++ )
      {
            for( nX = 0; nX < sizeBmp.cx; nX++ )
            {
                  COLORREF clrColored = bmpColored.GetPixel( nX, nY );
                  if(         clrTransparent != COLORREF(-1L)
                        &&    clrTransparent == clrColored
                        )
                        continue;
                  CExtBitmap::stat_RGBtoHSL( clrColored, &lfHue, &lfSaturation, &lfLuminance );
                  if( lfLuminance >= lfMinWhiteLuminance )
                        continue;
                  int nByteIndex = nX / 8;
                  BYTE & _byte = pCurrSurfacePtr[ nByteIndex ];
                  int nBitIndex = nX % 8;
                  nBitIndex = 7 - nBitIndex;
                  BYTE nMaskValue = BYTE( 1 << nBitIndex ); 
                  _byte &= ~nMaskValue;
            }
            pCurrSurfacePtr += nCountOfBytesPerLine;
      }
BITMAP bi;
    bi.bmType = 0;
    bi.bmWidth = sizeBmp.cx;
    bi.bmHeight = sizeBmp.cy;
    bi.bmWidthBytes = nCountOfBytesPerLine;
    bi.bmPlanes = 1;
    bi.bmBitsPixel = 1;
    bi.bmBits = pBinarySurface;
HBITMAP hBitmapMono = ::CreateBitmapIndirect( &bi );
    delete [] pBinarySurface;
    if( hBitmapMono == NULL )
      {
            ASSERT( FALSE );
            return NULL; // mono bitmap handle creation error
      }
      if( bDeleteColoredBitmapOnSuccess )
            ::DeleteObject( hBitmapColored );
      return hBitmapMono; // success
}

It is not difficult task to convert any BPP to any higher BPP. More difficult task is conversion to a lower BPP, especially when using bitmaps with indexed color format. The conversion into color BPP is called color quantization. It is performed by libraries like CxImage:

http://www.codeproject.com/bitmap/cximage.asp
http://www.codeproject.com/bitmap/cquantizer.asp

Hello,

many thanks four the code. I really appreciate your support. The code will be useful for our task.

One final question: is my assumption correct, that - regardless from the bmp format being loaded into CExtIconEditDlg - the format of the bitmap stored in CExtIconEditDlg::m_icon after returning from CExtIconEditDlg ::DoModal() is always 32 BPP ?

The format of the returned bitmap is the same as the format of the bitmap you started editing. The bitmap cloning during editing in the undo-redo buffer of image editor control is performed by the CExtBitmap::stat_CloneBitmap() method.

Hello,

I am afraid now, that we begin to go round in circles.

Once more:

Originally I wrote, that we do the following:

HBITMAP hbmpSourceMono; // given bitmap with 1 BPP

CExtIconEditDlg dlg( this );

dlg.m_icon.AssignFromHBITMAP(hbmpSourceMono, (COLORREF)-1);

if (dlg.DoModal() == IDOK)
{
...
}

with hbmpSourceMono being a monochrome bitmap with 1 BPP. I told you, that we have problems to extract a bitmap in the original 1BPP format from dlg.m_icon after calling DoModal(). Then you provided us with comprehensive source code, that fullfills this task. Finally you stated, that the format of the returned bitmap is the same as the format of the bitmap we started editing.

I am lost now. If the format of the returned bitmap is the same as the original bitmap (which was 1 BPP), why do we have those problems to extract a bitmap with 1 BPP and why do we need the comprehensive code, you provided us before???

We have checked the image editor’s code again. It seems it returns the same image format as assigned in all cases except 1 for BPP bitmaps. Please use the function we wrote for you for adjusting 1BPP bitmaps.

Hello,

understood, many thanks.

Your code will help to overcome this restriction. Do you plan to extend CExtIconEditDlg to make it treat all bitmap formats (including 1BPP consistently)?

Frankly speaking, we think 1BPP bitmaps with regard to Ui is relic of past. Even if you need some simple two color (black and white for instance) graphic images, you’d better use 32-bit images with alpha channel because they can contain very smooth edges of graphic shapes. It would be extremely interesting to know why and for which purposes you are using so low color images in your project?