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

Re: Change REFIID to IID* for stack declarations in oaidl_p.c to avoid initializers being discarded

by Mike Hearn

On Sun, 27 Jun 2004 21:13:17 +0200, Marcus Meissner wrote: >> + It is usually generated directly by MIDL, however this file has >> + been tweaked since then to account for slight differences in the way >> + gcc and MSVC++ compile it. In particular, in some functions REFIIDs >> + declared on the stack have been converted to plain IID* in order to eliminate >> + the constness of the REFIID type, ensuring that the zero initializer is not >> + discarded. >> + >> + Therefore, please do not regenerate this file. >> +*/ > > The comment is not understandable. > > Shouldn't it read '"REFIID" should not be used, because the value is not "const"'? I don't understand that wording of the comment either unfortunately :( It's not that it shouldn't be used, it's more that in the original generated code (I think) it was written like this: REFIID riid; (REFIID ) riid = 0; I'm not sure why MIDL did this, but as REFIID is simply #define REFIID const IID* const changing those instances to IID* should "fix" it without any warnings about assigning to a readonly variable. REFIID can still be used elsewhere though. thanks -mike

June 27, 2004

Re: Tooltips: Start of Balloon Support

by Filip Navara

Dmitry Timoshkov wrote: >"Robert Shearman" <R.J.Shearman(a)warwick.ac.uk> wrote: > >>This patch marks the start of balloon tip support. Note that it also has >>the same bug as native due to regions not being handled the same as on >>Windows. >> >> > >Do you mean a bug in FrameRgn? If yes, ReactOS might have that bug >fixed in their implementation. Could you have a look? > If that's the case, I'll try to fix it. I will propably have time to look at this issue tomorrow. Regards, Filip

June 27, 2004

Re: Change REFIID to IID* for stack declarations in oaidl_p.c to avoid initializers being discarded

by Marcus Meissner

On Sun, Jun 27, 2004 at 05:23:27PM +0100, Mike Hearn wrote: > This can cause a subtle marshalling failure in some cases > (inter-thread IDispatch, native MSI, etc) > > > Mike Hearn <mh(a)codeweavers.com> > Change REFIID to IID* for stack declarations in oaidl_p.c to avoid > initializers being discarded > Index: oaidl_p.c > =================================================================== > RCS file: /home/wine/wine/dlls/oleaut32/oaidl_p.c,v > retrieving revision 1.2 > diff -u -p -w -r1.2 oaidl_p.c > --- oaidl_p.c 5 Sep 2003 23:08:33 -0000 1.2 > +++ oaidl_p.c 27 Jun 2004 16:17:04 -0000 > @@ -1,5 +1,14 @@ > -/* this ALWAYS GENERATED file contains the proxy stub code */ > +/* This file contains the proxy/stub code for core COM interfaces. > > + It is usually generated directly by MIDL, however this file has > + been tweaked since then to account for slight differences in the way > + gcc and MSVC++ compile it. In particular, in some functions REFIIDs > + declared on the stack have been converted to plain IID* in order to eliminate > + the constness of the REFIID type, ensuring that the zero initializer is not > + discarded. > + > + Therefore, please do not regenerate this file. > +*/ The comment is not understandable. Shouldn't it read '"REFIID" should not be used, because the value is not "const"'? Ciao, Marcus

June 27, 2004

Re: Wine/WineX and subfs

by Marcus Meissner

On Sun, Jun 27, 2004 at 06:04:32PM +0200, Uwe Bonnes wrote: > Hallo, > > when trying to install programs from subfs mounted CDROMS, wine stumbled on > several occasions (GetVolumenInformation?), running on Suse 9.1. I had not > yet time to investigate deeper. However the announcement for WineX 3.3.2 > tells explicit about changes for CDROMs and subfs for Suse9.1. Anybody from > Transgaming reading hear, willing to submit a patch or explain the changes? The CD detection code is different between WineX and WineHQ. I tried both winefile and notepad (file->open) and both detect the subfs mounts correctly and fine. Ciao, Marcus

June 27, 2004

Re: FILE_GetNtStatus Converting errno 38 to STATUS_UNSUCCESSFUL

by Jürgen Leibner

"Jürgen Leibner" <juergen.leibner(a)t-online.de> schrieb im Newsbeitrag news:cbhj09$kct$1(a)sea.gmane.org... Hi there, > i'm trying to install a Software for Schools using wine-20040505-SuSELinux90.i586.rpm. Now, after updating to wine-20040615-SuSELinux90.i586.rpm, the Error doesn't apear anymore! The setup.exe does its job correct. After this the MS-Installer is invoked to do the rest of the job with an msi-file. He crashes at once. So i think i have to contact the programmer of the software to ask him what the installer wants to do at crashtime. For this i will dump the windump output to explain the point of crashing. Thanks to all, Jürgen

June 27, 2004

Wine/WineX and subfs

by Uwe Bonnes

Hallo, when trying to install programs from subfs mounted CDROMS, wine stumbled on several occasions (GetVolumenInformation?), running on Suse 9.1. I had not yet time to investigate deeper. However the announcement for WineX 3.3.2 tells explicit about changes for CDROMs and subfs for Suse9.1. Anybody from Transgaming reading hear, willing to submit a patch or explain the changes? Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

June 27, 2004

Re: Change Desktopsize for _INS5576._MP in config

by Uwe Bonnes

>>>>> "Uwe" == Uwe Bonnes <bon(a)elektron.ikp.physik.tu-darmstadt.de> writes: Uwe> Changelog: documentation/samples/config: Uwe> [AppDefaults\\_INS5576._MP\\x11drv] Change Desktop size to 800x600 Uwe> With a desktop siye of 640y480, Sage PC Kaufmann Setup aborts, Uwe> asking for higher resolution. Looking at http://frankscorner.org/modules.php?op=modload&name=Sections&file=index&req… the small desktopsize used until now for _INS5576._MP seems to cause problems with the photoshop installer too... -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

June 27, 2004

Re: Tooltips: Start of Balloon Support

by Dmitry Timoshkov

"Robert Shearman" <R.J.Shearman(a)warwick.ac.uk> wrote: > This patch marks the start of balloon tip support. Note that it also has > the same bug as native due to regions not being handled the same as on > Windows. Do you mean a bug in FrameRgn? If yes, ReactOS might have that bug fixed in their implementation. Could you have a look? -- Dmitry.

June 27, 2004

Suppressing NTDLL Relay messages

by Uwe Bonnes

Hallo, as the include and exclude lists for relay logging are initialized later than our ntdll is loader, relay logs contain all ntdll calls, even if the user didn't request them. Does anybody have an idea how to make the ntdll relay behave better? Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

June 27, 2004

Re: dbghelp crashes

by Eric Pouech

Jon Griffiths a écrit : > Hi, > > Some overly long stabs cause crashes when debug tracing is enabled. > > Cheers, > Jon > > +dlls/dbghelp/stabs.c > Prevent the debug buffer from overflowing on long stabs You shouldn't use debugstr_an with a fixed value, since all strings are '\0' terminated. Either fix dlls/ntdll/debugtools.c to handle properly those long string conditions (and remove the ugly call to abort()), or use min(256, strlen(ptr)) as the actual length of the string in debugstr_an calls. I'd prefer the first solution. A+

June 27, 2004