January 2006 - Wine-devel - list.winehq.org

Labels on rotated display not drawing in 0.9 release
by Phil Lodwick 06 Jan '06

06 Jan '06

Hi, I have a display in the portrait position. I am using Option "Rotate" "90" in my XF86Config-4 file. When I rotate the screen like this, labels on Menus, Buttons, Dialog Boxes, etc do not display until after an expose event (eg. running a cursor over them). I have tried Wine 0.9, Wine 0.9.4 and CrossOver 5.0. I use winemine to test Wine, and NotePad in the CrossOver Office distribution to test that configuration -- all fail. If I don't rotate OR if I revert to the 20050524 release of Wine, the problem goes away. Any suggestions on where to start the hunt? Thanks, Phil

1 0

Re: Wine 0.9.5 no longer runs Dragon Naturally Speaking
by Scott Ritchie 06 Jan '06

06 Jan '06

There's no apt repository for old packages, however old debs are available here: http://sourceforge.net/project/showfiles.php?group_id=6241&package_id=174803 I'll put a link to there on the download-deb page at winehq soon. I've forwarded this message to wine-devel as that's the proper place to report regressions, rather than emailing me. Thanks, Scott Ritchie On Fri, 2006-01-06 at 07:36 +0100, Emmanuel Charpentier wrote: > Dear Scott, > > As stated in the title, Wine 0.9.5 no longer runs Dragon Naturally > Speaking (voice recognition). I Cc this mail to peter(a)piments.com, > because he's seriously interested with this program (he published the > how-to-install document...). The program stops early in the setup > process (an error dialog box pops up *behind* DNS' flash screen) : the > error message refers to an "Access violation" error and points to > Dragon's own log (which reports a segment violation and barfs a somewhat > largish stack dump). > > Using DNS on a daily basis, I wish to revert to wine 0.9.4 or 0.9.3, but > I am unable to find a repository for the older Debian packages. Is there > a repository for these packages ? > > Sincerely yours, > > Emmanuel Charpentier

2 1

Re: OLEAUT32: Output a FIXME message only once.
by Cihan Altinay 06 Jan '06

06 Jan '06

Hi, Mike McCormack wrote: > > Unless somebody is planning to fix this soon, it would be nice to just > output it once to remind us that it needs to be fixed later. It's a > memory leak, so it's not worth frightening our users more than once over. > > Mike > > > ChangeLog: > Output a FIXME message only once. > > dlls/oleaut32/typelib.c | 7 ++++++- > 1 files changed, 6 insertions(+), 1 deletions(-) > > > ------------------------------------------------------------------------ > > a06a1b1d8dbb246eede84fd894e273641477d1fc > diff --git a/dlls/oleaut32/typelib.c b/dlls/oleaut32/typelib.c > index 4ef590e..70c1d1f 100644 > --- a/dlls/oleaut32/typelib.c > +++ b/dlls/oleaut32/typelib.c > @@ -4287,7 +4287,12 @@ static ULONG WINAPI ITypeInfo_fnRelease( > it means that function is called by ITypeLib2_Release */ > ITypeLib2_Release((ITypeLib2*)This->pTypeLib); > } else { > - FIXME("destroy child objects\n"); > + static int once = 0; > + if (once) I think it should read if (!once) :-) > + { > + once = 1; > + FIXME("destroy child objects\n"); > + } > > TRACE("destroying ITypeInfo(%p)\n",This); > if (This->Name) Cihan

1 0

Re: commctl32 datetime usability fixes
by James Hawkins 06 Jan '06

06 Jan '06

On 1/4/06, Siim Karus <siim04(a)ut.ee> wrote: > Hi! > > I modified datetime control to work with some Visual C++ wizard > generated applications. Also some usability fixes were made (typable > time, updown control notifications processing, calendar popup without > dropdown button disabled, style information synchronisation improved). Please send one fix/modification per patch and one patch per email. -- James Hawkins

1 0

wcmd & start
by Stefan Leichter 06 Jan '06

06 Jan '06

Hello, i found a problem during running a batch file. The problem is the start command. Win9x has it as separate comand (as we currently have it in wine). The Nt based Windows Versions have it as builtin comand in cmd. Wcmd does not know the start command because the start command ist installed in windows/comand directory, but the directory ist not in the path known by wcmd. An easy fix might be to add the windows/comand directory to the path variable on startup of wcmd. Is the fine for everyone? If not please suggest a better solution. Bye Stefan

5 4

Re: [comctl32] Reduce memory usage of the syslink control (Take 5)
by Robert Shearman 05 Jan '06

05 Jan '06

Thomas Weidenmueller wrote: >Reduce memory usage of the syslink control and fix dependent code. > > > Alexandre has already said what he doesn't like about the patch, but I'll be a bit more specific. Put the below changes into a separate patch plus the moving around of the LIF_STATE case in SYSLINK_SetItem, because they are really separate changes. >@@ -878,7 +879,7 @@ > if((Current->Type == slLink) && (Current->u.Link.state & LIS_FOCUSED) && infoPtr->HasFocus) > { > COLORREF PrevColor; >- PrevColor = SetBkColor(hdc, OldBkColor); >+ PrevColor = SetTextColor(hdc, OldBkColor); > DrawFocusRect(hdc, &bl->rc); > SetBkColor(hdc, PrevColor); > } >@@ -972,8 +973,7 @@ > /* clear the document */ > SYSLINK_ClearDoc(infoPtr); > >- textlen = lstrlenW(Text); >- if(Text == NULL || textlen == 0) >+ if(Text == NULL || (textlen = lstrlenW(Text)) == 0) > { > return TRUE; > } >@@ -984,8 +984,9 @@ > /* Render text position and word wrapping in memory */ > HDC hdc = GetDC(infoPtr->Self); > SYSLINK_Render(infoPtr, hdc); >- SYSLINK_Draw(infoPtr, hdc); > ReleaseDC(infoPtr->Self, hdc); >+ >+ InvalidateRect(infoPtr->Self, NULL, TRUE); > } > > return TRUE; >@@ -1531,11 +1553,8 @@ > case WM_SIZE: > { > HDC hdc = GetDC(infoPtr->Self); >- if(hdc != NULL) >- { >- SYSLINK_Render(infoPtr, hdc); >- ReleaseDC(infoPtr->Self, hdc); >- } >+ SYSLINK_Render(infoPtr, hdc); >+ ReleaseDC(infoPtr->Self, hdc); > return 0; > } > > > -- Rob Shearman

1 0

Re: gdi32 test
by Dmitry Timoshkov 05 Jan '06

05 Jan '06

"Stefan Leichter" <Stefan.Leichter(a)camLine.com> wrote: > ChangeLog > --------------- > add a test for CreateDCW 1. there is no need to import CreateDCW dynamically, it always exists 2. dlls/user/tests/monitor.c,test_enumdisplaydevices() test already does this, so your test is redundant. -- Dmitry.

3 3

Question regarding the Wine Vs WineLib performance
by Ananth M 05 Jan '06

05 Jan '06

Hi All, This question is related to the performance of wine using wine and WineLib I have a win32 program that uses a dll. I am checking the timing to execute a function that is exported by the dll I wanted to execute this program in Linux (using wine and WineLib) and check the performance. When I executed the win32 program in Linux using wine <win32program> The execution time of the function exported by the dll is as same as in direct execution of the program in windows Then I converted the dll into .so and compiled and linked this .so with win32 program using WineLib and Wineg++ The timing that was taken to execute the function exported by the dll ( through .so ) is almost two times, compread with executing using wine. Does any one has any reference for further investigation on this problem or any information Thank you Ajay

4 3

PATCH: assembler _invoke
by Marcus Meissner 05 Jan '06

05 Jan '06

On Wed, Jan 04, 2006 at 09:14:18PM -0800, Dan Kegel wrote: > This would definitely be better handled in assembly... Here is the first try. It leaves the old version for non-intel and non-gcc compilers. Ciao, Marcus Index: typelib.c =================================================================== RCS file: /home/wine/wine/dlls/oleaut32/typelib.c,v retrieving revision 1.200 diff -u -r1.200 typelib.c --- typelib.c 3 Jan 2006 12:42:34 -0000 1.200 +++ typelib.c 5 Jan 2006 09:07:25 -0000 @@ -4843,7 +4843,33 @@ switch (callconv) { case CC_STDCALL: +#if defined(__i386__) && defined(__GNUC__) + __asm__( + /* store all stuff into registers now, since we modify the stack */ + " mov %1, %%ecx\n" + " mov %2, %%edx\n" + " mov %3, %%esi\n" + " test %%ecx, %%ecx\n" + " jz 1f\n" /* no args, nothing to push */ + /* + * Arguments are pushed from right to left. + * We use intel memory move operations. + */ + " std\n" /* direction: backward */ + "2: lodsl\n" /* moves *esi -> eax, esi-= 4 */ + " pushl %%eax\n" + " dec %%ecx\n" + " jnz 2b\n" + "1: cld\n" /* direction: forward (default) */ + " call *%%edx\n" + " mov %%eax, %0\n" + : "=g" (res) + : "g" (nrargs), "g" (func), "g" (args + (nrargs-1)) + : "memory", "cc", "ax", "si", "cx", "dx" + ); + break; +#else switch (nrargs) { case 0: res = func(); @@ -4905,6 +4931,7 @@ break; } break; +#endif default: FIXME("unsupported calling convention %d\n",callconv); res = -1;

1 0

New, flex-related warnings
by Gerald Pfeifer 05 Jan '06

05 Jan '06

Using flex 2.5.4 and GCC 3.4, I started getting the following new warnings for my Wine builds about two days ago (which I never got before): ./ppl.l:176:1: warning: "YY_NO_UNPUT" redefined ./ppl.l:129:1: warning: this is the location of the previous definition ./parser.l:57:1: warning: "YY_NO_UNPUT" redefined warning: this is the location of the previous definition ./parser.l:58:1: warning: "YY_NO_TOP_STATE" redefined ./parser.l:23:1: warning: this is the location of the previous definition ./parser.l:116:1: warning: "YY_NO_UNPUT" redefined ./parser.l:100:1: warning: this is the location of the previous definition ./parser.l:117:1: warning: "YY_NO_TOP_STATE" redefined ./parser.l:101:1: warning: this is the location of the previous definition ./debug.l:73:1: warning: "YY_NO_UNPUT" redefined ./debug.l:22:1: warning: this is the location of the previous definition ./macro.lex.l:42:1: warning: "YY_NO_UNPUT" redefined ./macro.lex.l:22:1: warning: this is the location of the previous definition I'm afraid, this is caused by the following patchset: date: 2006-01-03 12:43:14 +0000; author: julliard; state: Exp; lines: +4 -2 Alexandre Julliard <julliard(a)winehq.org> Added some flex options to avoid compiler warnings. Moved options from the Makefile into the source. Gerald

1 0