October 2004 - Wine-Devel - list.winehq.org

wine developer guide
by Eric Pouech Oct. 20, 2004

Oct. 20, 2004

While looking at the WDG (and please don't count me in about arguing whether it's WDG or WD'sG or WD'G), I came onto the Advanced Topics book after the two big ones ("Developing Wine" and "Wine Architecture") I does seem very odd to me as it right now: - it just contains "random" bits of information on some parts of wine - we don't have (yet) a uniform way to qualify something of advanced So my proposal would be to: - get rid of "Advanced Topics" book altogether, and keep only "Developing Wine" (the way we'd like development going on) and "Wine architecture" (the way it works - or it's supposed to) - move the various items from "Advanced Topics" to where they belong (see below) - maybe tag in both the "developing" and "architecture" books which parts are really advanced (change of font, specific paragraph type... pick up the visual effect you like most) The mapping that could be done would be: 14.1 Keyboard => Architecture/X11DRV DLL 14.2 Undocumented API => somewhere in the "developing" chapter 14.3 accelerators => Architecture/USER32.DLL 14.4 HW trace => somewhere in the "developing" chapter (likely in debugging) 15 Porting to new OS => somewhere in the "developing" chapter (likely in writing patches, to be renamed as "coding practice" or so) 16 Consoles => Architecture/KERNEL32.DLL 17 Address space => Architecture/6.3.4 (as advanced material) 18 Regression testing => somewhere in the "developing" chapter (in conformance tests) Likewise, the 5.1 about adding new languages could be moved into the "coding practice" part. If noone objects, I'll submit the changes. A+

1 0

Stack overflows
by Robert Lunnon Oct. 20, 2004

Oct. 20, 2004

I have been trying to make wine work properly with Solaris xlib and have been experiencing a few stack overflow exceptions eg 0009:Ret kernel32._LeaveSysLevel() retval=00000000 ret=7f86a7ba 0009:Ret user32.USER_Unlock() retval=00000000 ret=7f53f837 0009:Call ntdll.RtlEnterCriticalSection(7f579840) ret=7f549249 0009:Ret ntdll.RtlEnterCriticalSection() retval=00000000 ret=7f549249 0009:Call ntdll.RtlLeaveCriticalSection(7f579840) ret=7f54926a 0009:Ret ntdll.RtlLeaveCriticalSection() retval=00000000 ret=7f54926a 0009:Call ntdll.RtlEnterCriticalSection(7f579840) ret=7f549249 0009:Ret ntdll.RtlEnterCriticalSection() retval=00000000 ret=7f549249 err:seh:setup_exception stack overflow 0 bytes in thread 0009 eip ca096003 esp 7fc11000 stack 7fc10000-7fd10000 As I recall PE executables specify the stack size in the PE header so in the past I have used a little windows utility under wine to increase the stack allocation to the program, sometimes with mixed results. I Thought I'd post here to find out if this is known problem and what the cause is. I had presumed it was probably due to the extra call layering through xlib adding too much overhead to the stack for the app. Assuming this is the issue there were two actions I was looking at taking 1. A Wine utility to adjust the stack allocation for the executable 2. Adjust the stack allocation in wine to allow a little extra for wine over and above that requested by the app. Opinions ? Bob

1 0

Reimplementing SHGetFolder functions
by Juan Lang Oct. 20, 2004

Oct. 20, 2004

Any reason this patch wasn't applied? http://www.winehq.org/hypermail/wine-patches/2004/10/0167.html I know it's big, but I hope the test cases help demonstrate its correctness. Let me know if there's anything else I can do to help a review. Thanks, --Juan __________________________________ Do you Yahoo!? Yahoo! Mail Address AutoComplete - You start. We finish. http://promotions.yahoo.com/new_mail

2 1

Re: Startmenu
by Robert Shearman Oct. 20, 2004

Oct. 20, 2004

Thorsten Kani wrote: > Nice Patch - looks good now! > i have changed the offsets and used SM_CXBORDER instead of SM_CXEDGE. > (According to MSDN, cxedge is used for 3D while cxborder is used for > Flat ) > This seems to fix the"draw below" issue. Visual Experience comes now > really near native. > > I noticed that SM_CXEDGE is used everywhere, regardless of Style > TBSTYLE_FLAT. > Please review my attached Patch. It seems to me that this Stylechecks > shoud be used everywhere. I don't think this is correct. If you change the return value for GetSystemMetrics(SM_CXEDGE) in wine/windows/sysmetrics.c then you can see that it does affect the spacing between border and icon (but our toolbar control doesn't exactly match the native for this). > As for String position issue, maybe i can help out with the controlspy > check. > Im interested in understanding the controls and wanted to play around > with CSpy anyway. > Suggestions ? You can download controlspy here: http://www.microsoft.com/msj/0798/controlspy.aspx It is very useful for throwing messages at the common controls, but it cannot test custom draw functionality and it is becoming dated (it is missing new functionality from comctl32 v5.80 and new controls). Rob

2 2

Re: dlls/x11drv/winpos.c::SWP_DoOwnedPopups()
by Dmitry Timoshkov Oct. 20, 2004

Oct. 20, 2004

"Peter Riocreux" <par+wine_patches(a)silistix.com> wrote: > The comment above SWP_DoOwnedPopups in dlls/x11drv/winpos.c suggests > that the only thing missing is what I *might* have fixed here, > therefore the patch also removes the FIXME and the WARN. It compiles > and doesn't break any behaviour I can test, but I am only guessing > whether I have done the right thing. You have to write a test case and add it to existing ones in dlls/user/tests/win.c to confirm that the patch is correct. -- Dmitry.

2 3

DrawTextA: regarding the nCount variable
by William Poetra Yoga H Oct. 20, 2004

Oct. 20, 2004

In the MSDN, it is stated that if nCount is 1, then it is assumed that lpString is a null-terminated string. But in the DrawTextExA function, it is -1, as in the code: if (count == -1) count = strlen(str); Shouldn't this be 1 instead of -1? CMIIW. (But theForger's tutorial also uses -1) _______________________________ Do you Yahoo!? Declare Yourself - Register online to vote today! http://vote.yahoo.com

3 3

Re: ADVAPI32: implement ChangeServiceConfigA using ChangeServiceConfigW (take 2)
by Robert Shearman Oct. 20, 2004

Oct. 20, 2004

Mike McCormack wrote: > > oops. The macro functions for duplicating strings already exist, so > use them instead. > > Mike > > > ChangeLog: > * implement ChangeServiceConfigA using ChangeServiceConfigW > >------------------------------------------------------------------------ > >Index: dlls/advapi32/service.c >=================================================================== >RCS file: /home/wine/wine/dlls/advapi32/service.c,v >retrieving revision 1.60 >diff -u -r1.60 service.c >--- dlls/advapi32/service.c 27 Sep 2004 20:35:27 -0000 1.60 >+++ dlls/advapi32/service.c 20 Oct 2004 07:28:54 -0000 >@@ -810,7 +810,8 @@ > > static inline VOID SERV_free( LPWSTR wstr ) > { >- HeapFree( GetProcessHeap(), 0, wstr ); >+ if (wstr) >+ HeapFree( GetProcessHeap(), 0, wstr ); > } > > There is no need for this check. Freeing NULL pointers is not an error and is handled by the RtlFreeHeap code. Rob

1 0

Re: fix tbn_deletingbutton
by Dmitry Timoshkov Oct. 19, 2004

Oct. 19, 2004

"Thorsten Kani" <beebix(a)gmx.net> wrote: > I'm a bit unsure about my solution, but actually everything works. > -complaints appreciated- > --- toolbar.c 23 Sep 2004 22:51:14 -0000 1.194 > +++ toolbar.c 10 Oct 2004 17:29:01 -0000 > @@ -3191,12 +3191,20 @@ > TOOLBAR_INFO *infoPtr = TOOLBAR_GetInfoPtr (hwnd); > INT nIndex = (INT)wParam; > NMTOOLBARW nmtb; > + WCHAR Buffer[256]; > > if ((nIndex < 0) || (nIndex >= infoPtr->nNumButtons)) > return FALSE; > > memset(&nmtb, 0, sizeof(nmtb)); > nmtb.iItem = nIndex; > + memcpy(&nmtb.tbButton, &infoPtr->buttons[nIndex], sizeof(nmtb.tbButton) ); > + nmtb.cchText = 256; > + nmtb.pszText = Buffer; > +#if (_WIN32_IE >= 0x500) > + nmtb.rcButton =(RECT)infoPtr->buttons[nIndex].rect; > +#endif > + > TOOLBAR_SendNotify((NMHDR *)&nmtb, infoPtr, TBN_DELETINGBUTTON); 1. do not use _WIN32_IE version checks in Wine code, they have no sense 2. there is no point in setting a pointer to a not initialized buffer 3. there is no need to cast RECT to a RECT -- Dmitry.

3 8

Re: add epoll implementation to libwine_port (take 3)
by Chris Dodd Oct. 19, 2004

Oct. 19, 2004

Mike McCormack <mike(a)codeweavers.com> wrote in news:4174D681.9050503(a)codeweavers.com: > A few more changes after consulting with Alexandre. > > ChangeLog: > * add epoll implementation to libwine_port : > + __asm__ __volatile__( "pushl %%ebx; movl %2,%%ebx; int $0x80; popl > %%ebx" + : "=a" (ret) : "0" : A random question here from someone who is not too familiar with wine internals -- why are you using inline asm code instead of syscall(2)? It seems like syscall would be both simpler and more portable. Chris Dodd cdodd(a)acm.org

1 0

Re: add epoll implementation to libwine_port (take 3)
by Boaz Harrosh Oct. 19, 2004

Oct. 19, 2004

Mike McCormack wrote: > > > ChangeLog: > * add epoll implementation to libwine_port > >.... > > >+ __asm__ __volatile__( "pushl %%ebx; movl %2,%%ebx; int $0x80; popl %%ebx" > > Grate ! [Q] Does that mean that we directly check if it is supported in kernel, bypassing glibc all together? how do we avoid conflicts with glibc?

4 5