- Wine-Devel - list.winehq.org

RE: More MSVC fixes
by Patrik Stridvall May 5, 2002

May 5, 2002

> How are you planning on linking the dlls/resources once wine is up 2 > date on MSVC? No specific plan. I just fix issues that I can find acceptable solutions to one by one in hope that all will be fixed eventually. You and Alexandre have fixed some of them that I wasn't sure how to solve in an acceptable way so, as I said, perhaps we will solve them all eventually. :-) > Can MSVC build wrc resources and link them? The formats are not 100% compatible. Specificly WRC supports embedded icons and bitmaps. Microsoft RC requires a link to the binary. However Alexandre doesn't like binary files in the CVS for good reasons so making the RC files 100% Microsoft compatable wouldn't be a good idea either. Not sure how to solve this in a good way other than actually using WRC on Windows of course.

6 8

Issues building the spec.o on win32 with mingw
by Steven Edwards May 4, 2002

May 4, 2002

I get this on almost every *.spec.o Administrator(a)LAPTOP /cygdrive/d/src/winehq/wine/dlls/commdlg $ make PATH="../../unicode:$PATH" ../../tools/wrc/wrc -I. -I. -I../../include -I../../ include -o rsrc.res -r rsrc.rc PATH="../../unicode:$PATH" ../../tools/winebuild/winebuild -L../../dlls -o comm dlg.spec.c -spec commdlg.spec gcc -mno-cygwin -c -I. -I. -I../../include -I../../include -D__MINGW__ -D_WINDO WS -DWINE_NOWINSOCK -Wall -mpreferred-stack-boundary=2 -D__WINE__ -D_REENTRANT -o commdlg.spec.o commdlg.spec.c /cygdrive/c/DOCUME~1/ADMINI~1/LOCALS~1/Temp/ccU3cEQ7.s: Assembler messages: /cygdrive/c/DOCUME~1/ADMINI~1/LOCALS~1/Temp/ccU3cEQ7.s:1104: Warning: unknown se ction attribute 'a' /cygdrive/c/DOCUME~1/ADMINI~1/LOCALS~1/Temp/ccU3cEQ7.s:1106: Error: unknown pseu do-op: `.previous' make: *** [commdlg.spec.o] Error 1 Any ideas? I really don't need the spec because I can do a -l if I can build the .a for ntdll and a few others. I can use dlltool with the *.defs and then I only need winebuild to make the *.defs. Only problem is winebuild wants so's not dll's when building a def or looking for imports. I was going to have a look at the loader/winedump code for ideas but I thought I would ask here first. Thanks Steven "Every revolution was once a thought in one man's mind" - Ralph Waldo Emerson

4 8

A dsound regression
by Duane Clark May 4, 2002

May 4, 2002

A little more regression testing... The 5/1 dsound patch causes this on one of my apps: err:module:BUILTIN32_LoadLibraryExA loaded .so but dll sound.dll still not found - library environment problem or version conflict, check your setup. err:module:BUILTIN32_LoadLibraryExA loaded .so but dll sound.dll still not found - library environment problem or version conflict, check your setup. err:wave:DSDB_MapPrimary (0x4568ca04): Could not map sound device for direct access (Input/output error)

2 1

Bug 446: SetFileAttributes and BufferOverflow
by Ann and Jason Edmeades May 4, 2002

May 4, 2002

Looking at Bugzilla bug 446 and the msdn, it would appear (and testing confirms) that: rc = SetFileAttributes("fff", FILE_ATTRIBUTE_NORMAL | FILE_ATTRIBUTE_READONLY ); _is_ valid and sets readonly. In the msdn "However, all other values override FILE_ATTRIBUTE_NORMAL." The wine code (files\file.c) prints a fixme if anything is set alongside attribute_normal: if (attributes & FILE_ATTRIBUTE_NORMAL) { attributes &= ~FILE_ATTRIBUTE_NORMAL; if (attributes) FIXME("(%s):%lx illegal combination with FILE_ATTRIBUTE_NORMAL.\n", lpFileName,attributes); } I would suggest a trace point rather than a fixme or remove this check altogether As to the buffer overflow, the only thought I have is a missing \n on the end of the error msg: if (!S_ISDIR(buf.st_mode)) FIXME("SetFileAttributes expected the file '%s' to be a directory", lpFileName); Thoughts? I can submit a patch for these simple issues, but I cant explain the other issues (re files expected to be directories) Regards, Jason

1 0

Application Database RFC
by Tony Lambregts May 4, 2002

May 4, 2002

I have done a lot of thinking about the Application Database http://appdb.codeweavers.com lately and how the rating system works. We currently have 848 Applications in there but only 58 with a 5 star rating which works out to less than 10% and thats the good news. The bad news is that of those 58 only 26 rate that without windows installed. After takeing a closer look, none of the apps I looked at actualy rated 5 star by my way of judging. Now maybe I'm overly critical but in my way of thinking If the app doesn't install in wine then it shouldn't get a 5. IMHO We need to set criteria that give us a better/less subjective way of measuring how well we are doing. So this is what I have come up with. 1. Installation (total possible 2) Installs in wine with fake windows 2 points Installs in wine with windows mounted 1 point Does not install in wine 0 points 2. Running & DLLS (total possible 2) Runs fine using all builtin dlls 2 points Runs fine using native dlls 1 point Does not run/crashes on startup 0 points 3 Running general (total possible 2) program does not crash ever 2 points Program crashes only after long use or rare aspect used 1 point program crashes on startup or doing common function 0 points 4. Screen Presentation/layout (total possible 2) App looks exactly like it should 2 points Small problems that do not affect working with app 1 point App is unusable because of display problems 0 points 5. Fonts (total possible 2) Fonts look exacty right 2 points Fonts not exactly right but acceptable 1 point Fonts unreadable or totaly wrong. 0 points 6. Mouse (total possible 2) Mouse behaves as expected 2 points Mouse behaves strange but can still use app 1 point Mouse behaves badly and is unusable 0 points 7. Keyboard (total possible 2) Keyboard reacts as expected. 2 points Some keys/key combinations dont work but most do. 1 point Keyboard locks up or unresposive 0 points 8. Files save/restore (total possible 2) Able to save and restore files without any problems 2 points Can save and restore files but have problems 1 points Cannot save or restore files 0 points 9. Sound (total possible 2) Not applicable Sound is correct 2 points sound is ok most of the time but has minor glichs 1 point Sound is garbled or missing 0 points 10.Joystick/gamepad (total possible 2) Not applicable joystick/gamepad works correctly 2 points joystick/gamepad have problems but mostly work 1 point joystick/gamepad unresponsive 0 points 11.Registy/rememberd state (total possible 2) Not applicable Program remembers key settings from one session to next 2 points Program remembers most settings but not all 1 point program does not remember most settings at all 0 points 12. Date tested. Verison of wine used Each of the sections should have a comment field and a place to enter a url to a bug report if applicable Applications could be scored based as a percentage. Applications that use all critera would need all 22 points to get a 100% rating. For other applications some of these criteria are not applicable and their score would use a lower base score. Idealy we would be able to look at the AppDB and list all programs that have problems with any of the criteria Probably the most usefull aspect of this is that it gives users and developers a real way of measuring how well wine really is doing and where problems that affect the most programs are. I have cross posted this because I think this is important to both users and developers. Any comments, criticisims or helpfull suggestions welcome Tony Lambregts

9 15

Re: xrender patch
by Duane Clark May 4, 2002

May 4, 2002

Jeremey White wrote: >> That does fix it, though I was getting used to the client side fonts :( >> What version of XFree86 is needed to get a good xrender? I am running >> version 4.1.0. > > You need 4.2.0. You can also just build libXrender.so, > which is not that hard to pull out of CVS and build by itself. > It turns out that two other applications (that worked prior to 4/23) fail even with the Huw patch that disables xrender. They give: err:ntdll:RtlpWaitForCriticalSection section 0x40b724a4 "x11drv_main.c: X11DRV_CritSection" wait timed out, retrying (60 sec) fs=023f err:ntdll:RtlpWaitForCriticalSection section 0x407a9718 "user.c: USER_SysLevel" wait timed out, retrying (60 sec) fs=008f wine: Unhandled exception, starting debugger... But I never actually get a debugger prompt, so I don't have more info then that. In any case, I just downloaded the binary tar file: http://ftp.xfree86.org/pub/XFree86/4.2.0/binaries/Linux-ix86-glibc22/Xbin.t… and took out just the libXrender.so.1.1, and installed it. Everything seems to work fine with that.

1 0

RE: FW: PATCH - check for ftruncate/fix for makedep when missing ftruncate (mingw)
by Steven Edwards May 4, 2002

May 4, 2002

> Steven, > > patches should go to wine-patches prefered. > Yes I know, I have had auto compleate on and sometimes my mailing lists get mixed up. I will keep a closer eye on where/when my stuff gets sent out so theres not so much noise. Thanks Steven

1 0

Send PATCH - check for ftruncate/fix for makedep when missing ftruncate (mingw) to /dev/null
by Steven Edwards May 3, 2002

May 3, 2002

I hit send when I was trying to save. I will be sending the new patch tonight. Thanks Steven "Every revolution was once a thought in one man's mind" - Ralph Waldo Emerson > -----Original Message----- > From: Steven Edwards [mailto:Steven_Ed4153(a)yahoo.com] > Sent: Friday, May 03, 2002 6:53 PM > To: 'wine-devel(a)winehq.com' > Subject: FW: PATCH - check for ftruncate/fix for makedep when > missing ftruncate (mingw) > > > Changelog: More Mingw work. > configure, configure.ac - Configure check for > ftruncate. > configure, configure.ac - Configure check for chsize > port.h - Use chsize instead of ftruncate if present. > port.h - Disable use of HAVE_PREAD and > HAVE_WRIT on Mingw > makedep.c - patch to use port.h > > Fix to disable > > "Every revolution was once a thought in one man's mind" > - Ralph Waldo Emerson >

1 0

FW: PATCH - check for ftruncate/fix for makedep when missing ftruncate (mingw)
by Steven Edwards May 3, 2002

May 3, 2002

Changelog: More Mingw work. configure, configure.ac - Configure check for ftruncate. configure, configure.ac - Configure check for chsize port.h - Use chsize instead of ftruncate if present. port.h - Disable use of HAVE_PREAD and HAVE_WRIT on Mingw makedep.c - patch to use port.h Fix to disable "Every revolution was once a thought in one man's mind" - Ralph Waldo Emerson

1 0

RE: fix brainfade
by Medland, Bill May 3, 2002

May 3, 2002

Alexandre Yes, go with Francois' Bill > -----Original Message----- > From: Francois Gouget [mailto:fgouget(a)free.fr] > Sent: Wednesday, May 01, 2002 3:27 PM > To: Wine patches (E-mail) > Subject: Re: fix brainfade > > > > Or (but mine may have space/tab problems): > > Index: wine/dlls/comctl32/status.c > =================================================================== > RCS file: /home/wine/wine/dlls/comctl32/status.c,v > retrieving revision 1.43 > diff -u -r1.43 status.c > --- wine/dlls/comctl32/status.c 29 Apr 2002 17:12:01 -0000 1.43 > +++ wine/dlls/comctl32/status.c 1 May 2002 21:03:46 -0000 > @@ -636,7 +636,7 @@ > if (nPart == 0x00ff) { > part = &infoPtr->part0; > } else { > - if (infoPtr->parts && (nPart < 0 || nPart >= > infoPtr->numParts)) { > + if (infoPtr->parts && nPart >= 0 && nPart < > infoPtr->numParts)) { > part = &infoPtr->parts[nPart]; > } > } > > > > -- > Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ > The greatest programming project of all took six days; on > the seventh day the > programmer rested. We've been trying to debug the *&^%$#@ > thing ever since. > Moral: design before you implement. > > >

3 2