- Wine-devel - list.winehq.org

Re: some other playsound fixes
by Alexandre Julliard 29 May '02

29 May '02

Eric Pouech <eric.pouech(a)wanadoo.fr> writes: > typedef struct tagWINE_PLAYSOUND { > - HANDLE hThread; > - HANDLE hReadyEvent; > - BOOL bStop; > - LPCWSTR pszSound; > - HMODULE hMod; > - DWORD fdwSound; > - BOOL bLoop; > - BOOL bAlloc; > + volatile BOOL bLoop : 1, > + bAlloc : 1; > + LPCWSTR pszSound; > + HMODULE hMod; > + DWORD fdwSound; > + struct tagWINE_PLAYSOUND* lpNext; > } WINE_PLAYSOUND, *LPWINE_PLAYSOUND; If you need to make variables volatile, something is wrong with the synchronization, and volatile is the wrong fix (especially on a bit field where updates can't be atomic). What is this supposed to fix? -- Alexandre Julliard julliard(a)winehq.com

2 1

Second current CVS compile failure
by Guy L. Albertelli 29 May '02

29 May '02

After bypassing previous error, got this one: gcc -c -I. -I. -I../../include -I../../include -g -O2 -Wall -fPIC -D__WINE_ _ -D_REENTRANT -I/usr/X11R6/include -o information.o information.c ld -r comcat_main.o factory.o manager.o register.o information.o -o comcat.dll.tmp.o strip --strip-unneeded comcat.dll.tmp.o LD_LIBRARY_PATH="../../library:../../unicode:$LD_LIBRARY_PATH" ../../tools/winebuild/winebuild -fPIC -D__WINE__ -sym comcat.dll.tmp.o -o comcat.spec.c -spec ./comcat.spec -L../../dlls -lole32 -ladvapi32 -lkernel32 gcc -c -I. -I. -I../../include -I../../include -g -O2 -Wall -fPIC -D__WINE_ _ -D_REENTRANT -I/usr/X11R6/include -o comcat.spec.o comcat.spec.c LD_LIBRARY_PATH="../../library:../../unicode:$LD_LIBRARY_PATH" ../../tools/winebuild/winebuild -fPIC -D__WINE__ -o comcat.dll.dbg.c -debug -C. comcat_main.c factory.c manager.c register.c information.c gcc -c -I. -I. -I../../include -I../../include -g -O2 -Wall -fPIC -D__WINE_ _ -D_REENTRANT -I/usr/X11R6/include -o comcat.dll.dbg.o comcat.dll.dbg.c gcc -shared -Wl,-Bsymbolic comcat.spec.o comcat_main.o factory.o manager.o register.o information.o comcat.dll.dbg.o -o omcat.dll.so -L../../dlls -L../../library -lwine -L../../ole -lwine_uuid -l m information.o: In function `COMCAT_GetCategoryDesc': /home/albertel/t1/wine/dlls/comcat/information.c:392: undefined reference to `swprintf' collect2: ld returned 1 exit status make[2]: *** [comcat.dll.so] Error 1 Guy Albertelli <<galberte(a)neo.lrun.com>>

2 1

RFC - patch for psapi and missing imports on Mingw target
by Steven Edwards 29 May '02

29 May '02

This kind of a dirty hack but just a idea and it shouldn't have a effect on cygwin as it will use the win32api package also. Mabey a better name then MINGW_IMPORTS can be used. This patch allows psapi.dll to be (almost) built. Mingws libkernel32.a doesn't contain the import information for the toolhelp APIs. Instead it models Visual Studio 6 and implents them in libth32.a. Does anyone know if this has changed in Visual Studio .NET? If so then I will submit a patch to mingw for the Win32api. If not, then I belive using libth32.a is the most proper way. Would a better solution be for me to make the make implib rule and just use wines libkernel32.a? I havent really looked at doing it because NTDLL, MSVCRT,GDI,USER and KERNEL32 are not targets of my port and may contain exports lacking on a Windows host. The only bug after this patch is - psapi_main.o(.text+0x1248):psapi_main.c: undefined reference to `wine_server_cal l' Thanks Steven "Every revolution was once a thought in one man's mind" - Ralph Waldo Emerson cvs diff -u Makedll.rules.in Index: Makedll.rules.in =================================================================== RCS file: /home/wine/wine/dlls/Makedll.rules.in,v retrieving revision 1.27 diff -u -r1.27 Makedll.rules.in --- Makedll.rules.in 22 May 2002 21:32:49 -0000 1.27 +++ Makedll.rules.in 29 May 2002 05:32:10 -0000 @@ -35,7 +35,7 @@ # Rules for .dll files $(MODULE): $(OBJS) $(MODULE).dbg.o $(SPEC_DEF) Makefile.in - $(DLLWRAP) $(DLLWRAPFLAGS) --def $(SPEC_DEF) --implib $(MODULE:.dll=.a) -o $@ $(OBJS) $(MODULE).dbg.o -L$(DLLDIR) $(DELAYIMPORTS:%=-l%) $(IMPORTS:%=-l%) $(ALL_LIBS) + $(DLLWRAP) $(DLLWRAPFLAGS) --def $(SPEC_DEF) --implib $(MODULE:.dll=.a) -o $@ $(OBJS) $(MODULE).dbg.o -L$(DLLDIR) $(DELAYIMPORTS:%=-l%) $(IMPORTS:%=-l%) $(MINGW_IMPORTS) $(ALL_LIBS) $(SPEC_DEF): $(WINEBUILD) cvs diff -u Makefile.in Index: Makefile.in =================================================================== RCS file: /home/wine/wine/dlls/psapi/Makefile.in,v retrieving revision 1.11 diff -u -r1.11 Makefile.in --- Makefile.in 11 May 2002 23:06:34 -0000 1.11 +++ Makefile.in 29 May 2002 05:33:48 -0000 @@ -4,6 +4,7 @@ VPATH = @srcdir@ MODULE = psapi.dll IMPORTS = kernel32 ntdll +MINGW_IMPORTS = -lth32 LDDLLFLAGS = @LDDLLFLAGS@ SYMBOLFILE = $(MODULE).tmp.o

1 0

Current CVS compile failure
by Guy L. Albertelli 29 May '02

29 May '02

Using the current CVS I get the following errors compiling Cdrom.c. My Linux version is a pretty stock 2.2.12 (yes it is old). Have we terminated support for older versions?? gcc -c -I. -I. -I../../include -I../../include -g -O2 -Wall -fPIC -D__WINE_ _ -D_NTSYSTEM_ -DBINDIR="\"/usr/local/bin\"" -DETCDIR="\"/usr/local/etc\"" -D_REENTRANT -I/usr/X11R6/include -o cdrom.o cdrom.c cdrom.c:83: `CDROM_PACKET_SIZE' undeclared here (not in a function) cdrom.c: In function `CDROM_GetIdeInterface': cdrom.c:119: `SG_EMULATED_HOST' undeclared (first use in this function) cdrom.c:119: (Each undeclared identifier is reported only once cdrom.c:119: for each function it appears in.) cdrom.c: In function `CDROM_ScsiPassThroughDirect': cdrom.c:1071: storage size of `sense' isn't known cdrom.c:1097: `CGC_DATA_WRITE' undeclared (first use in this function) cdrom.c:1100: `CGC_DATA_READ' undeclared (first use in this function) cdrom.c:1103: `CGC_DATA_NONE' undeclared (first use in this function) cdrom.c:1109: `CDROM_SEND_PACKET' undeclared (first use in this function) cdrom.c:1071: warning: unused variable `sense' cdrom.c: In function `CDROM_ScsiPassThrough': cdrom.c:1134: storage size of `sense' isn't known cdrom.c:1167: `CGC_DATA_WRITE' undeclared (first use in this function) cdrom.c:1170: `CGC_DATA_READ' undeclared (first use in this function) cdrom.c:1173: `CGC_DATA_NONE' undeclared (first use in this function) cdrom.c:1179: `CDROM_SEND_PACKET' undeclared (first use in this function) cdrom.c:1134: warning: unused variable `sense' make[2]: *** [cdrom.o] Error 1 I did a search and some of these symbols didn't show up till 2.2.16 or later. Guy Albertelli <<galberte(a)neo.lrun.com>>

1 0

Re: Font patch
by Dmitry Timoshkov 29 May '02

29 May '02

"Dustin Navea" <speeddymon(a)yahoo.com> wrote: > This basically just adds Aliases for some of MS's > Default Fonts to the wine default config... fixes bug > #721... There is no need to add aliases to the config file. Config file should provide replacements for basic system fonts only, common aliases are handled by graphics/x11drv/xfont.c, XFONT_LoadAliases(). In the case of the client-side rendered fonts, there is some font aliasing code too. Probably it should be enhanced though. -- Dmitry.

1 0

Re: [Bug] #724 found! Someone needs to fix it?
by Andreas Mohr 28 May '02

28 May '02

On Tue, May 28, 2002 at 04:08:46PM -0500, Boris Reisig wrote: > Ive been downloading the latest CVS of wine and I have noticed that when doing a clean install of wine on a drive without windows, I have noticed that wine couldnt install Word '97 viewer, Office '97 setup, Excel '97 viewer.[Make sure you have a clean temp directory too] I have done *alot* of CVS regression thanks to Andriy Palamarchuk. This was the following error. > > err:font:ReadFontDir Can't open directory "/c/windows/Fonts" > err:module:PE_fixup_imports Modules (file) MSSETUP.dll (which is needed by E:\~MSSETUP.T\~msstfqf.t\acmsetup.exe) not found. > > So I have tracked it down to wine/files/profile.c May 11,2002 17:59:41 CDT. [http://www.winehq.com/hypermail/wine-cvs/2002/05/0092.html] > > If I CVS before that date/time, The applications work fine. Hope this helps someone to possible fix it? ---------------------------- revision 1.67 date: 2002/05/11 22:59:41; author: julliard; state: Exp; lines: +11 -5 Andreas Mohr <andi(a)rhlx01.fht-esslingen.de> Make PROFILE_GetString account for default value in all cases. ---------------------------- Sounds like it's me then :-\ Argl. I really wonder how this could have broken it... Maybe via some unrelated function that uses the same profile function that I patched... Run wine --debugmsg +relay,+profile MyPrg.exe &>/tmp/wine.log before and after that patch , bzip2 or gzip wine.log and mail them to me. Thanks ! -- Andreas Mohr Stauferstr. 6, D-71272 Renningen, Germany Tel. +49 7159 800604 http://mohr.de.tt

1 0

dinput keyboard patch
by arjen＠blehq.org 28 May '02

28 May '02

Why is my patch not committed?

1 0

Request for small build system change
by Steven Edwards 28 May '02

28 May '02

When I use cygwin or msys to build wine/tools I need to link to wrc and wmc to libberty.a. I've tried to write a little fix for it but havent had much luck and still have a few other things I am working on. Thanks Steven "Every revolution was once a thought in one man's mind" - Ralph Waldo Emerson

1 0

CoGetClassObject and multiple servers requested
by Medland, Bill 28 May '02

28 May '02

(I guess this is for Marcus) I thought I had heard something about Local Server COM being in place now but when I went to try it with our product I saw no attempt to do it (error message about CLSCTX_LOCAL_SERVER registered but not supported). So I looked at CoGetClassObject and I see that the Remote_CoGetClassObject (which I presume is attempting to do the LOCAL_SERVER) is only called if a local server is requested and an inproc server or handler is not. Is there a good reason why the Remote_CoGetClassObject is not called on failure to detect a registered inproc server or handler (which the existing structure of the function supported)? Bill

2 1

Why start a COM local server twice?
by Medland, Bill 28 May '02

28 May '02

Is this an oversight or a deliberate policy? Remote_CoGetClassObject creates the local server process and hands over to create_marshalled_proxy If create_marshalled_proxy can't immediately find the pipe then it will create the process again. Is it safe to drop the first one? How about the second? It looks to me as if Remote_CoGetClassObject is the only place that calls create_marshalled_proxy so I presume we can remove either. Bill

2 1