August 2005 - Wine-Devel - list.winehq.org

Re: X11: Problems with ConfigureNotify events that are received/processed too late

by Michael Kaufmann

Hi all, Some other propositions on how to fix this: 1. Don't handle X events in MsgWaitForMultipleObjectsEx, just wait for them. 2. Don't cache the X11 window rectangle and always use XGetGeometry to get the current window rectangle. That will surely solve the problem, but also calls like GetClientRect would always need a call to the X server, making things less efficient. The first proposal should not be too hard to implement for anyone familiar with Xlib. I hope somebody can do it. Regards Michael > Hi all, > > I'm trying to get the dialog boxes of Word 95 working. They often > appear with the minimal window size (about 70x50 pixels), so they're > unusable. With KDE's window manager (KWin) this happens very often. > With Enlightenment and GNOME's window manager (Metacity), it only > happens from time to time. > > I've looked for the reason, and I think it's very closely related to > WINE bug 1265 which Duane Clark has tried to fix here: > http://www.winehq.com/hypermail/wine-patches/2003/02/0095.html > > That's what's going on (CVS version of WINE, managed window mode): > > 1. Word creates a dialog with an initial position of (0,0) and size > 0x0 (using CreateWindowEx) > 2. Word calls SetWindowPos to set the correct dialog position and size > 3. Word calls SetWindowPos to show the dialog (SWP_SHOWWINDOW) > 4. The dialog gets the focus, the main window is being deactivated > 5. In the WM_NCACTIVATE procedure of the main window, Word calls > GetQueueStatus. WINE will process X events. > 6. WINE receives a ConfigureEvent for the initial dialog position and > size (0x0) and saves this to data->whole_rect in > X11DRV_sync_window_position > (7. If you're lucky, WINE also receives a ConfigureEvent for the > correct dialog position) > 8. Wine shows the dialog and maps it. It calls X11DRV_set_wm_hints, > which calls set_size_hints. This procedure uses data->whole_rect as > the dialog's rectangle. But this stored rectangle is completely wrong > because WINE has not yet received all ConfigureEvents. > 9. The window manager has to resize the dialog to the wrong size. > > > The best solution would be to process all X events before setting the > window position in SetWindowPos, but I don't know if that's possible. > > I've tried another approach: WINE should not set the window size hints > when mapping a window (step 8.). I've attached a patch, but it's not > 100% correct, because the style of a window (e.g. the WS_THICKFRAME > style) may be changed by an application using SetWindowLong and > SetWindowPos (with the SWP_FRAMECHANGED flag). > > WINE doesn't look for the SWP_FRAMECHANGED flag and updates the X11 > window every time if it's visible or gets visible. So if an > application calls SetWindowPos with the SWP_SHOWWINDOW flag, WINE > needs to set the size hints because the window has been hidden and its > style may have been changed. > > What would be the correct fix for this problem? Any ideas? > > Thanks, > Michael

Aug. 22, 2005

Re: Throwing in an idea (probably it was discussed before though)

by John Smith

>Because it's a tedious and boring task to narrow down those unknown bugs in >closed-source apps. And that's exactly why we ask you (since you got access >to the sources) to tell us what the application is trying to do which >doesn't work in Wine... Ahem. And how long it usually takes to fix the bug for not-top-10 application? And please, don't suggest to fix it ourselves - it is not going to happen in corporate environment. As for 'tell us what the application is trying to do' - you've got all APIs in your hands, so what's the problem to track down system calls? The issue our app is experiencing is quite a subtle one - it works Ok with Managed=N or with Desktop=YYYxZZZ (or from CrossOver BTW), but with default WineHQ distribution when it tries to minimize one of it's top-level windows with an explicit call to ShowWindow, the window is not only minimized, but also reduced to 1x1 pixel; when restoring it, sometimes it is restored normally, sometimes not. This behaviour was reported for Gnome/KDE/Xfce and can be fixed with 'Managed' or 'Desktop' settings I've mentioned above. >>(we checked relevant portions of our code though for potential bugs and >>found nothing) >Even if your application relies on undocumented behaviour ("potential bugs" >heh) it should work - if it doesn't that's a bug in Wine which should be >fixed. Well, in fact WINE compatibility wasn't our goal in such a check (WINE installation base is still negligible, and probably will remain negligible until some attention will be paid to out-of-the-box compatibility - at least like in CrossOver); it was just a good reason to check for potential incompatibilities under 'native' Win platforms. Also note that having WINE work perfectly is not our goal (personally I would like it, but the company coudln't care less). >Managed=N breaks focusing for everything on my box (bug #2149)... As it would fix the problem for 99% of our users, and won't affect those who doesn't run our app, IMO it is still good enough. Also I remember you've just told that fixing bugs is easy - why does it still exist then? _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today - it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Aug. 22, 2005

Dragon ns8 problem

by Jean Magnan de Bornier

Hello all, My machine: Linux-gentoo, wine 20050725 I'm quite new to wine, just installed it in order to use dns8, of which I have the french version. I looked at recent threads on this list and have been more or less hopeful, but right now I have had no success; here is the log of my last attempt with the command "wine setup.exe": === Début de l'écriture dans le journal : 22/08/2005 14:18:02 === Début de l'action 14:18:02 : INSTALL. Début de l'action 14:18:02 : ISMsiServerStartup. Fin de l'action 14:18:03 : ISMsiServerStartup. Valeur renvoyée : 1. 1: The InstallScript engine version currently installed on this machine is adequate. Début de l'action 14:18:03 : ISStartup. 1: Event 'E183186' is created 1: GetInstallDriver, Can not find InstallDriver in ROT table, Return Code = 0x800401e3 1: {DDDD0C4B-57F7-4A85-ACF0-DB3FC8F1DBB4} 1: ISMsiServerStartup Failure. Unknown Exception. Fin de l'action 14:18:03 : ISMsiServerStartup. Valeur renvoyée : 1603. 1: ISStartUp Failure. OpenEvent, Error = 0x2 Fin de l'action 14:18:03 : ISStartup. Valeur renvoyée : 3. Début de l'action 14:18:03 : SetupCompleteError. Informations 2898. Tahoma8, Tahoma, 0 Informations 2898. TahomaBold10, Tahoma, 0 Début de l'action 14:18:06 : ISCleanUpFatalExit. Fin de l'action 14:18:06 : ISCleanUpFatalExit. Valeur renvoyée : 3. DEBUG: Error 2896: Executing action ISCleanUpFatalExit failed. Erreur interne 2896. ISCleanUpFatalExit Fin de l'action 14:18:06 : SetupCompleteError. Valeur renvoyée : 3. Fin de l'action 14:18:06 : INSTALL. Valeur renvoyée : 3. === Fin de l'écriture dans le journal : 22/08/2005 14:18:06 === MSI (c) (0C:0E): Produit : Dragon NaturallySpeaking 8 -- L'installation a échoué. The eighth line of this seems to be tne interesting one; what would this InstallDriver and this ROT table be? Can I find them somewhere? Any hope still? TIA, -- Jean Magnan de Bornier | Cours Victor Hugo e-mots: jean at bornier.net | 13980 Alleins France T 08 70 39 34 03 | P 06 09 17 35 87

Aug. 22, 2005

(no subject)

by John Smith

>You can also set it yourselves in in your app. As configuration >is in registry, every app can modify it. I like this idea, thanks. There are still 2 issues with such a soultion: a) most likely, it won't work until 2nd start of the application (we will modify it on the first run, and it most likely won't be applied until the second run); b) current WINE guides are discouraging us from detecting WINE - and messing up with Windows Registry, especially HKEY_LOCAL_MACHINE, on all the millions of non-WINE machines is not a thing I would really like to do (the last thing we want is to be caught by some antivirus, which monitors activity it considers suspicious and changing HKEY_LOCAL_MACHINE can easily qualify). _________________________________________________________________ Dont just search. Find. Check out the new MSN Search! http://search.msn.click-url.com/go/onm00200636ave/direct/01/

Aug. 22, 2005

Re: Throwing in an idea (probably it was discussed before though)

by Felix Nawothnig

John Smith wrote: >> This will just allow developers to hide bugs in wine and slow >> development even further, I thing it's a bad idea. > Ahem. To say it politely - I didn't get an impression that WINE > developers are desperately looking for unknown bugs to fix. Because it's a tedious and boring task to narrow down those unknown bugs in closed-source apps. And that's exactly why we ask you (since you got access to the sources) to tell us what the application is trying to do which doesn't work in Wine... Actually fixing bugs is usually easy, tracking them down OTOH... > (we checked relevant portions of our code though for potential bugs and > found nothing) Even if your application relies on undocumented behaviour ("potential bugs" heh) it should work - if it doesn't that's a bug in Wine which should be fixed. And by the way: Managed=N breaks focusing for everything on my box (bug #2149)... :-) Felix

Aug. 22, 2005

Re: Throwing in an idea (probably it was discussed before though)

by John Smith

>Then just tell your users to set that in winecfg, AFAIK winecfg allows app >specific settings. 1. It is still not 'out-of-the-box' - and from this point of view it doesn't matter much whether it is hacking config file or using GUI; 80% of end-users will try it and throw it away if it doesn't work without hacking settings; 20% of others will ask questions, and will hack config or run winecfg, but those 80% will be lost. 2. (not really relevant to this discussion, but relevant to the whole issue of supporting end-users): binary RPMs for RedHat/Fedora on winehq.org are still 20050524, and winecfg is pretty much useless there. If WINE team does care about end-users, it would be nice to have binary RPMs for the most popular distribution not that much out-of-date. >I'm sure your great guys but any such mechanism could be easily abuse by >lazy programmers, Don't see much ways to abuse in general - eventually WINE is supposed to run _any_ Win application, right? Also I've suggested that config/winecfg per-application settings should override those 'hints', so user is still in control of it. >also once it wasn't needed some sort of backwards compatibility may even be >needed, If all it is about is overriding currently existing settings, no backward compatibility is required. >One simple solution would be to provide default settings for apps that have >known problems, Alexandre hasn't ruled that out if it helps some apps to >work out of the box. Ok, this probably is even better; what about integrating such a thing with AppDB and automatically include some special field from AppDB data into winehq distribuitions? _________________________________________________________________ Dont just search. Find. Check out the new MSN Search! http://search.msn.click-url.com/go/onm00200636ave/direct/01/

Aug. 22, 2005

Re: Throwing in an idea (probably it was discussed before though)

by Ivan Leo Puoti

John Smith wrote: > Situation is pretty simple. We have an application, which works Ok under > WINE, provided that Managed=N specified Then just tell your users to set that in winecfg, AFAIK winecfg allows app specific settings. I'm sure your great guys but any such mechanism could be easily abuse by lazy programmers, also once it wasn't needed some sort of backwards compatibility may even be needed, I don't think it's a great idea. One simple solution would be to provide default settings for apps that have known problems, Alexandre hasn't ruled that out if it helps some apps to work out of the box. Ivan.

Aug. 22, 2005

Re: X11DRV: more gcc 4.0 warning fixes

by Alexandre Julliard

Mike McCormack <mike(a)codeweavers.com> writes: > Index: dlls/x11drv/event.c > =================================================================== > RCS file: /home/mike/src/wine-cvs/wine/dlls/x11drv/event.c,v > retrieving revision 1.58 > diff -u -p -r1.58 event.c > --- dlls/x11drv/event.c 16 Aug 2005 16:02:45 -0000 1.58 > +++ dlls/x11drv/event.c 17 Aug 2005 01:09:41 -0000 > @@ -698,7 +698,7 @@ static void EVENT_DropFromOffiX( HWND hW > > if( !aux_long && p_data) /* don't bother if > 64K */ > { > - signed char *p = (signed char*) p_data; > + char *p = (char*) p_data; > char *p_drop; > > aux_long = 0; The code is storing -1 in there so it has to be a signed char (though that algorithm is really a nasty hack that should be fixed properly). > --- dlls/x11drv/text.c 23 Dec 2004 20:31:56 -0000 1.15 > +++ dlls/x11drv/text.c 16 Aug 2005 01:53:53 -0000 > @@ -390,7 +390,7 @@ X11DRV_ExtTextOut( X11DRV_PDEVICE *physD > wine_tsx11_lock(); > if (lfUnderline) > { > - long linePos, lineWidth; > + unsigned long linePos, lineWidth; > > if (!XGetFontProperty( font, XA_UNDERLINE_POSITION, &linePos )) > linePos = descent - 1; > @@ -405,7 +405,7 @@ X11DRV_ExtTextOut( X11DRV_PDEVICE *physD > } > if (lfStrikeOut) > { > - long lineAscent, lineDescent; > + unsigned long lineAscent, lineDescent; > if (!XGetFontProperty( font, XA_STRIKEOUT_ASCENT, &lineAscent )) > lineAscent = ascent / 2; > if (!XGetFontProperty( font, XA_STRIKEOUT_DESCENT, &lineDescent )) I think these should remain signed too, especially since lineDescent is set to -lineAscent. -- Alexandre Julliard julliard(a)winehq.org

Aug. 22, 2005

Re: [winedbg] Don't ever pass a NONCONTINUABLE exception

by Alexandre Julliard

Glenn Wurster <gwurster(a)scs.carleton.ca> writes: > Changelog: > Don't pass on non-continuable exceptions. Why do you want to do that? -- Alexandre Julliard julliard(a)winehq.org

Aug. 22, 2005

Re: MSXML: add some node tests

by Alexandre Julliard

Mike McCormack <mike(a)codeweavers.com> writes: > diff -u -p -r1.1 domdoc.c > --- dlls/msxml3/tests/domdoc.c 12 Aug 2005 11:25:05 -0000 1.1 > +++ dlls/msxml3/tests/domdoc.c 20 Aug 2005 07:38:47 -0000 > @@ -27,59 +27,65 @@ > #include "xmldom.h" > #include <stdio.h> > > +#ifndef STANDALONE > #include "wine/test.h" > +#else > +#define ok(cond,str) do{ if(!(cond)) fprintf(stderr,"line %d: %s",__LINE__,str); }while (0) > +#endif Please don't add that sort of thing again, simply copy wine/test.h over to Windows and build with -DSTANDALONE. -- Alexandre Julliard julliard(a)winehq.org

Aug. 22, 2005