March 2006 - Wine-Devel - list.winehq.org

GL_STENCIL-related build failures in dlls/wined3d/device.c
by Gerald Pfeifer March 11, 2006

March 11, 2006

The recent changes to dlls/wined3d/device.c cause the following build failures for me on FreeBSD 5.4, though I assume this is in fact not related to FreeBSD but to xorg 6.8.2 or libglut 6.4.1 on this system. device.c: In function `IWineD3DDeviceImpl_SetRenderState': device.c:3776: error: `GL_STENCIL_BACK_FAIL' undeclared (first use in this function) device.c:3776: error: (Each undeclared identifier is reported only once device.c:3776: error: for each function it appears in.) device.c:3777: error: `GL_STENCIL_BACK_PASS_DEPTH_FAIL' undeclared (first use in this function) device.c:3778: error: `GL_STENCIL_BACK_PASS_DEPTH_PASS' undeclared (first use in this function) gmake[2]: *** [device.o] Error 1 gmake[2]: Leaving directory `/.amd_mnt/nashira/files5/test/wine/dlls/wined3d' gmake[1]: *** [wined3d] Error 2 The most straightforward patch would be the one below, but it leaves three "variable may be used uninitialized" warnings in the first of the two affected spots. Gerald Index: device.c =================================================================== RCS file: /home/wine/wine/dlls/wined3d/device.c,v retrieving revision 1.145 diff -u -3 -p -r1.145 device.c --- device.c 4 Mar 2006 17:10:57 -0000 1.145 +++ device.c 5 Mar 2006 08:35:52 -0000 @@ -3226,9 +3226,15 @@ HRESULT WINAPI IWineD3DDeviceImpl_SetRen GLint action = StencilOp(Value); +#ifdef GL_STENCIL_FAIL glGetIntegerv(GL_STENCIL_FAIL, &stencilFail); +#endif +#ifdef GL_STENCIL_PASS_DEPTH_FAIL glGetIntegerv(GL_STENCIL_PASS_DEPTH_FAIL, &depthFail); +#endif +#ifdef GL_STENCIL_PASS_DEPTH_PASS glGetIntegerv(GL_STENCIL_PASS_DEPTH_PASS, &stencilPass); +#endif if(WINED3DRS_STENCILFAIL == State) { stencilFail = action; @@ -3773,9 +3779,15 @@ HRESULT WINAPI IWineD3DDeviceImpl_SetRen GLint action = StencilOp(Value); +#ifdef GL_STENCIL_BACK_FAIL glGetIntegerv(GL_STENCIL_BACK_FAIL, &stencilFail); +#endif +#ifdef GL_STENCIL_BACK_PASS_DEPTH_FAIL glGetIntegerv(GL_STENCIL_BACK_PASS_DEPTH_FAIL, &depthFail); +#endif +#ifdef GL_STENCIL_BACK_PASS_DEPTH_PASS glGetIntegerv(GL_STENCIL_BACK_PASS_DEPTH_PASS, &stencilPass); +#endif if(WINED3DRS_CCW_STENCILFAIL == State) { stencilFail = action;

1 1

AlphaImageLoader - tranparent pngs 4 IE
by Axel Huizinga March 10, 2006

March 10, 2006

Hi! Did anybody get DXImageTransform.AlphaImageLoader working to render transparent png in IE on wine? Cordially, Axel

4 3

Re: winegdb: Fix crash in empty first command
by Eric Pouech March 10, 2006

March 10, 2006

Vitor T. Sessak wrote: > 2006/3/7, Eric Pouech > >> >> 2006/3/6, Vitor T. Sessak <vitor.sessak at m4x.org >> <http://www.winehq.org/mailman/listinfo/wine-patches>>: >> >>> / > > />>/ Changelog: > />>/ winegdb: Fix crash in empty first command > />>/ > />>/ Winedbg crashes when pressing enter as its first command. > /> > >> >> this is wrong as the last_line will be HeapFree'd when the next >> command is >> keyed in. > > > You are right. I've changed it and fixed the FIXME too. your new code leaks the lines (you never free them) A+ -- Eric Pouech

1 1

Re: riched20: Implements FR_WHOLEWORD for EM_FINDTEXT
by Alexandre Julliard March 10, 2006

March 10, 2006

Lei Zhang <leiz(a)ucla.edu> writes: > --- dlls/riched20/editstr.h 2 Mar 2006 11:18:19 -0000 1.27 > +++ dlls/riched20/editstr.h 5 Mar 2006 06:36:50 -0000 > @@ -29,6 +29,7 @@ > #include <stdarg.h> > #include <stdio.h> > #include <stdlib.h> > +#include <wctype.h> You can't use anything from wctype.h in Wine. Use the functions from wine/unicode.h instead. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: WINED3D: Color fixups for vertex shaders
by Roderick Colenbrander March 10, 2006

March 10, 2006

> Direct3D stores vertex color data in BGRA order, OpenGL stores it in > RGBA order. This causes red and blue to be swapped in d3d applications > when run under wine. This patch fixes this problem for vertex shaders > by enabling the flag that causes red and blue to be swapped in > vshader_program_add_input_param_swizzle, when appropriate. > > Changelog: > - Enable color fixups for vertex shaders > In my local tree I have some simple d3d8 shaders working. (See for instance http://roderick.student.utwente.nl/dolphin1.png and dolphin2.png on windows; there's some color issue but apart from that it works). This patch causes in combination with the other vshader patch you sent some drawing issues. Check http://roderick.student.utwente.nl/dolphin3.png Regards, Roderick -- "Feel free" mit GMX FreeMail! Monat f�r Monat 10 FreeSMS inklusive! http://www.gmx.net

2 5

DDraw->WineD3D refcounting hack
by Stefan Dösinger March 10, 2006

March 10, 2006

Hello, I have a a question regarding some code that might be rejected by Alexandre when I submit it. The problem is that I have to do an refcount hack in my new ddraw code to destroy textures properly. I CC this mail to AJ because it's him who has the last word ;) WineD3D and Direct3D7 have different ideas about handling mipmap textures. In d3d7 they are complex surfaces - a bunch of surfaces with a root and sublevels. All surfaces in the compound have their own refcount, but they are all destroyed when the root is destroyed. In WineD3D there's the texture as a container, and the surfaces in the container have their refcount linked to the container(See the patch H. Verbeet sent a few days ago). In d3d7 the application destroys the first surface to release the whole texture. My first idea was to Release the WineD3DTexture. It would then call the Release method of all the sublevel ddraw surfaces, which would destroy them. This works for a correctly written application, but applications call GetAttachedSurface for the sublevels, which addrefs, and many do not Release the sublevels after GetAttachedSurface(). This leaves the sublevel surfaces with a refcount of 2 or more when the root is destroyed. To make sure that they are destroyed, I set their refcount to 1 when the root is destroyed. When the last reference to the WineD3D Texture is released, wineD3D will destroy the sublevel surfaces. Is such a thing acceptable for Wine? Does anyone have a better solution? If you need some code to look at, I'll upload an updated patch to my homepage later. Thanks, Stefan

2 2

Trying to fix bug 2400 - delete and comma
by agaro agaro March 10, 2006

March 10, 2006

Hi all, I'm trying to fix bug 2400 - removes one char and puts comma simultaneously. My guess: this old bug is in dlls/x11drv/keyboard.c or dlls/user/message.c, something about dead keys implementation. Can somebody help me with a little explanation or documentation? I don't know what scancode, keycode and vkey mean. Thanks, agaro

4 4

Re: Use a property for the colour dialog's information.
by Robert Shearman March 9, 2006

March 9, 2006

Mike McCormack wrote: >@@ -1217,7 +1220,7 @@ static INT_PTR CALLBACK ColorDlgProc( HW > DeleteDC(lpp->hdcMem); > DeleteObject(lpp->hbmMem); > HeapFree(GetProcessHeap(), 0, lpp); >- SetWindowLongPtrW(hDlg, DWLP_USER, 0); /* we don't need it anymore */ >+ SetPropW( hDlg, szColourDialogProp, lpp ); > > SetPropW( hDlg, szColourDialogProp, lpp ) should be SetPropW( hDlg, szColourDialogProp, NULL ). -- Rob Shearman

2 1

Re: dinput: Create single thread for mouse and keyboard hook procs. [try 3]
by Vitaliy Margolen March 9, 2006

March 9, 2006

Any problems with this patch? I have not received any additional comments. And it does fix at least two games in question (see bug reports). And I have not seen any side-affects from it either. Vitaliy Margolen Wednesday, February 22, 2006, 8:10:16 AM, Vitaliy Margolen wrote: > ChangeLog: > dinput: Create single thread for mouse and keyboard hooks. > Put keyboard & mouse hook callbacks into separate thread. > Move few global variables into object. Delete no longer used crit section. > For hooks to work properly hook callback have to be in a thread > with message loop. Some games create separate threads just to handle > mouse and/or keyboard events that do not have message loop. This patch fixes > such games (Bug 4264) - mouse and (Bug 2981) - keyboard. > Try#2 (based on suggestions from Alexandre): > Using a message window in a dedicated thread with normal message loop. > Hooks are installed and removed using SendMessage to this window. > Only one mouse and one keyboard hook is permitted. > Hook handles are managed by dedicated thread and not given back to > caller. > Try#3 (with Robs suggestions) > Use critical section. > dlls/dinput/device.c | 132 > ++++++++++++++++++++++++++++++++++++++++++ > dlls/dinput/device_private.h | 2 + > dlls/dinput/keyboard.c | 130 > ++++++++++++++++------------------------- > dlls/dinput/mouse.c | 33 ++++------- > 4 files changed, 195 insertions(+), 102 deletions(-)

1 0

Re: x11drv: Fix one case of "glXChooseFBConfig returns NULL" problem.
by Vitaliy Margolen March 9, 2006

March 9, 2006

Anything wrong with this patch? According to MSDN ChoosePixelFormat should pick closest mode, but not fail. Vitaliy Margolen Saturday, March 4, 2006, 10:27:11 PM, Vitaliy Margolen wrote: > ChangeLog: > x11drv: Fix one case of "glXChooseFBConfig returns NULL" problem. Using alpha bits >> 8 makes this function fail, while it should not. > dlls/x11drv/opengl.c | 3 ++- > 1 files changed, 2 insertions(+), 1 deletions(-)

1 0