Re: user32: WM_SETFONT on button shouldn't repaint directly

[Clinton Stimpson]

Anything wrong with this patch?  It hasn't been accepted.

Thanks,
Clint

Clinton Stimpson wrote:
> This patch includes a couple improvements over the patch submitted 
> yesterday.
>
> Fix WM_SETFONT on a button to invalidate the rect instead of painting it 
> directly.
> Test included.
>
> Thanks,
> Clinton Stimpson
>
> ChangeLog:
>    user32: don't repaint button directly when setting font
> Index: dlls/user32/button.c
>
> RCS file: /home/wine/wine/dlls/user32/button.c,v
> retrieving revision 1.1
> diff -u -r1.1 button.c
> --- dlls/user32/button.c	16 Nov 2006 14:02:17 -0000	1.1
> +++ dlls/user32/button.c	19 Jan 2007 06:03:45 -0000
> @@ -410,7 +410,7 @@
>  case WM_SETFONT:
>      set_button_font( hWnd, (HFONT)wParam );
> -
>    if (lParam) paint_button( hWnd, btn_type, ODA_DRAWENTIRE );
> -
>    if (LOWORD(lParam)) InvalidateRect(hWnd, NULL, TRUE);
>    break;
>
> case WM_GETFONT:
> Index: dlls/user32/tests/msg.c
>
> RCS file: /home/wine/wine/dlls/user32/tests/msg.c,v
> retrieving revision 1.15
> diff -u -r1.15 msg.c
> --- dlls/user32/tests/msg.c	9 Jan 2007 12:11:42 -0000	1.15
> +++ dlls/user32/tests/msg.c	19 Jan 2007 06:03:45 -0000
> @@ -4159,6 +4159,14 @@
>      { WM_CAPTURECHANGED, sent|wparam|defwinproc, 0 },
>      { 0 }
>  };
> +static const struct message WmSetFontButtonSeq[] =
> +{
> - { WM_SETFONT, sent },
> - { WM_PAINT, sent },
> - { WM_ERASEBKGND, sent|defwinproc },
> - { WM_CTLCOLORBTN, sent|defwinproc },
> - { 0 }
>
> +};
>
> static WNDPROC old_button_proc;
>
> @@ -4239,6 +4247,9 @@
>      unsigned int i;
>      HWND hwnd;
>      DWORD dlg_code;
> -
>
> MSG msg;
> -
>
> LOGFONTA lf;
> -
>
> HFONT zfont;
>
> subclass_button();
>
> @@ -4278,6 +4289,25 @@
>  SendMessageA(hwnd, WM_LBUTTONUP, 0, 0);
>  ok_sequence(WmLButtonUpSeq, "WM_LBUTTONUP on a button", FALSE);
> -
> - while (PeekMessage( &msg, 0, 0, 0, PM_REMOVE )) DispatchMessage( &msg );
> - flush_sequence();
> -
> - lstrcpyA(lf.lfFaceName, "Symbol");
> - lf.lfHeight = 10;
> - lf.lfItalic = 0;
> - lf.lfEscapement = 0;
> - lf.lfOrientation = 0;
> - lf.lfUnderline = 0;
> - lf.lfStrikeOut = 0;
> - lf.lfWeight = 3;
> - lf.lfWidth = 10;
> - zfont = CreateFontIndirectA(&lf);
> - SendMessageA(hwnd, WM_SETFONT, (WPARAM)zfont, TRUE);
> - while (PeekMessage( &msg, 0, 0, 0, PM_REMOVE )) DispatchMessage( &msg );
> - ok_sequence(WmSetFontButtonSeq, "WM_SETFONT on a button", FALSE);
> - DeleteObject(zfont);
> - DestroyWindow(hwnd);
>
> }

user32: WM_SETFONT on button shouldn't repaint directly

[Alexandre Julliard]

Clinton Stimpson  writes:
> Anything wrong with this patch?  It hasn't been accepted.

I suspect you need an UpdateWindow() too. To confirm this (or not) you
should avoid the PeekMessage loop in your test.