May 2012 - Wine-Devel - list.winehq.org

Re: [PATCH 1/3] shell32: Add IQueryInfo in ShellFolder2_GetUIObjectOf [try 2]
by Nikolay Sivov May 20, 2012

May 20, 2012

On 5/20/2012 18:47, Detlef Riekenberg wrote: > +HRESULT IQueryInfo_Constructor(LPCITEMIDLIST root, LPCITEMIDLIST item, void ** ppvOut) > +{ > + IQueryInfoImpl *qi = HeapAlloc(GetProcessHeap(), 0, sizeof(IQueryInfoImpl)); > + LPITEMIDLIST pidl = ILCombine(root, item); > + > + if (!pidl || !qi) { > + SHFree(pidl); > + HeapFree(GetProcessHeap(), 0, qi); > + return E_OUTOFMEMORY; > + } > + > + qi->IQueryInfo_iface.lpVtbl =&IQueryInfo_vtbl; > + qi->ref = 1; > + qi->pidl = ILClone(pidl); > + > + SHFree(pidl); I don't see a reason to ILClone() here. You could just store ILCombine() result.

1 0

Re: dsound: Fixed compilation error caused by 942f2ce7
by Jacek Caban May 19, 2012

May 19, 2012

Hi Alexander, On 5/19/12 3:35 PM, Alexander E. Patrakov wrote: > There is no dummy union u anymore. > This doesn't seem right. They are still present, just generated by widl. Jacek

2 2

re: [appdb] Applications working flawlessly using patched wine should be rated Gold
by Dan Kegel May 18, 2012

May 18, 2012

Frédéric Delanoy wrote: -Application works flawlessly with some DLL overrides, other settings or third party software. +Application works flawlessly with some DLL overrides, some patches applied, other settings or third party software. Really? IMHO they should still be silver. Patches are very hard for the average user to deploy without a third party front end like POL, and appdb is not about POL. - Dan

7 13

Re: [1/2] comctl32/tests: Added tests for treeview right mouse click handling (try 3)
by Marvin May 18, 2012

May 18, 2012

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=18480 Your paranoid android. === WNT4WSSP6 (32 bit treeview) === Timeout

1 0

Re: [PATCH 1/5] programs: First version of the joystick testing tool. Just lists connected joysticks.
by Alexandre Julliard May 18, 2012

May 18, 2012

Lucas Fialho Zawacki <lfzawacki(a)gmail.com> writes: > From: Lucas Fialho Zawacki <lfzawacki(a)gmail.com> > > --- > configure.ac | 1 + > programs/joystick/Makefile.in | 17 +++++++ > programs/joystick/main.c | 102 +++++++++++++++++++++++++++++++++++++++++ > 3 files changed, 120 insertions(+) > create mode 100644 programs/joystick/Makefile.in > create mode 100644 programs/joystick/main.c I don't think it makes sense to add a command-line application for this. If there are really things that the user needs to configure, it should be a control panel. But ideally we should retrieve the settings from the Unix environment. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: quartz: COM cleanup for Parser_OutputPin
by Michael Stefaniuc May 18, 2012

May 18, 2012

Hello Aric, On 05/18/2012 04:27 PM, Aric Stewart wrote: > diff --git a/dlls/quartz/avisplit.c b/dlls/quartz/avisplit.c > index 5027e90..bac1b24 100644 > --- a/dlls/quartz/avisplit.c > +++ b/dlls/quartz/avisplit.c > @@ -288,7 +288,7 @@ static HRESULT AVISplitter_next_request(AVISplitterImpl *This, DWORD streamnumbe > > static HRESULT AVISplitter_Receive(AVISplitterImpl *This, IMediaSample *sample, DWORD streamnumber) > { > - Parser_OutputPin *pin = (Parser_OutputPin *)This->Parser.ppPins[1+streamnumber]; > + Parser_OutputPin *pin = impl_Parser_OutputPin_from_IPin(This->Parser.ppPins[1+streamnumber]); this isn't right. impl_from_... should be used only from methods of the respective implementation. You can either: - Make This->Parser.ppPins an array of Parser_OutputPin instead of IPin - Or if you need to support application provided IPin then you'll need an unsafe_impl_from_... > HRESULT hr; > LONGLONG start, stop, rtstart, rtstop; > StreamData *stream = &This->streams[streamnumber]; bye michael

2 1

Logging issue ?
by GOUJON Alexandre May 18, 2012

May 18, 2012

Hi, To understand why a game does not work with the disc while it does with the ISO (created from the same disc), I made a log with +tid,+relay,+volume in both cases and noticed : - some lines (call or ret) are not printed / missing - some lines are half-printed and joined with the following I know sometimes a trailing \n is missing in TRACE() so the output is uncertain. But that doesn't seem to be the case here. Maybe it's not thread safe or something like that ? I tried using taskset 1 wine .. to force using one core but no luck. Here a small part of my log (order not modified, I just remove lines from other threads): 0030:Call KERNEL32.GetVolumeInformationA(09732288 "D:",097321ec,0000007f,00000000,00000000,097321d8,00000000,00000000) ret=0057b045 0030:trace:volume:GetVolumeInformationW L"\\??\\D:\\": found fs type 50009:Call wined3d.wined3d_mutex_unlock() ret=7ed6e3a6 0030:Ret KERNEL32.CreateFileA() retval=000001f4 ret=00c8b5ab 0030:Call KERNEL32.DeviceIoControl(000001f4,00041018,00000000,00000000,09732190,00000008,097321a4,00000000) ret=00c8b64a The game calls GetVolumeInformationA .. the trace is mixed with output of another thread .. I guess the game open the device (no call line) .. and then do some ioctl. No trace of how the game opened the device.. I can do several logs and try to compare each other to have a complete puzzle but that's not very handy. Any thought ? idea ? advice ? (I'm using packaged wine on x86_64) Thanks

2 1

Fwd: Re: Logging issue ?
by GOUJON Alexandre May 18, 2012

May 18, 2012

Also replying to wine-devel -------- Original Message -------- Subject: Re: Logging issue ? Date: Fri, 18 May 2012 16:52:34 +0200 From: GOUJON Alexandre <ale.goujon(a)gmail.com> To: Nikolay Sivov <bunglehead(a)gmail.com> On 05/18/2012 05:31 PM, Nikolay Sivov wrote: > On 5/18/2012 17:23, GOUJON Alexandre wrote: >> Any thought ? idea ? advice ? > I believe there was a discussion about that in the past, It usually > helps to use a ">>" redirect that appends to log file instead of > replacing its contents. But you'll obviously need to clean a log file > before starting anything Using>> instead of> seems to work. It's weird but if it works... Thanks for the advice !

1 0

Re: [PATCH 3/4] shell32: Add IQueryInfo to more shellfolder
by Nikolay Sivov May 18, 2012

May 18, 2012

On 5/18/2012 14:06, Detlef Riekenberg wrote: > + } else if (IsEqualIID(riid,&IID_IQueryInfo)&& (cidl == 1)) { > + pidl = ILCombine (This->pidlRoot, apidl[0]); > + pObj = (IUnknown *) IQueryInfo_Constructor (pidl); > + SHFree (pidl); > + hr = S_OK; > } else { Patch doesn't have a cast here, and looks really redundant. If you really need that it's better to make IQueryInfo_Constructor look more like class factory call - with HRESULT and void** parameter. Also it looks like you always need to combine pidls and cleanup after that, I think it's better to combine them in constructor and keep result in QueryInfo impl. (instead of making a clone). So constructor could look like "HRESULT constructor(root, pidl, pObj)".

1 0

Re: [PATCH 1/4] shell32: Add IQueryInfo in ShellFolder2_GetUIObjectOf
by Nikolay Sivov May 18, 2012

May 18, 2012

On 5/18/2012 14:06, Detlef Riekenberg wrote: > + if (ppv == NULL) > + return E_POINTER; This is usually redundant. > + > + if (IsEqualGUID(&IID_IUnknown, riid) || > + IsEqualGUID(&IID_IQueryInfo, riid)) { > + *ppv =&This->IQueryInfo_iface; > + IUnknown_AddRef((IUnknown*)*ppv); > + TRACE("returning IQueryInfo: %p\n", *ppv); > + return S_OK; > + } Instead of using IUnknown for some reason a clean way could be to call IQueryInfo_AddRef(iface). Trace is not needed here - you have a trace on top of that and FIXME trace on a failure, so you'll know. > + > + *ppv = NULL; > + FIXME("(%p)->(%s %p) interface not supported\n", This, debugstr_guid(riid), ppv); > + return E_NOINTERFACE; > +} > +IQueryInfo *IQueryInfo_Constructor(LPCITEMIDLIST pidl) > +{ > + IQueryInfoImpl *qi; > + > + TRACE("(%p)\n", pidl); > + > + qi = HeapAlloc(GetProcessHeap(), 0, sizeof(IQueryInfoImpl)); > + qi->IQueryInfo_iface.lpVtbl =&IQueryInfo_vtbl; > + qi->ref=1; > + qi->pidl=ILClone(pidl); > + > + TRACE("==> %p (pidl: %p)\n", qi, qi->pidl); > + return&qi->IQueryInfo_iface; > +} This doesn't handle HeaoAlloc() or ILClone failures.

1 0

Installing 64 bit visual c++ 2010 runtime?
by Dan Kegel May 17, 2012

May 17, 2012

I'd like to try a real 64 bit app, so on a 64 bit Ubuntu 12.04, I tried sudo apt-get install wine rm -rf ~/.wine export WINEARCH=win64 wget http://download.microsoft.com/download/3/2/2/3224B87F-CFA0-4E70-BDA3-3DE650… wine vcredist_x64 and that installs fine... but only puts anything in c:\windows\syswow64, and the 64 bit app I'm trying is rather upset that it can't find mfc100.dll. The log it writes into c:\users\$LOGNAME\*.html says (if you look at the source, or click 'verbose' in a web browser) "platform not supported". This happens even if I use winecfg to set win7 mode. Happily, it looks like one can just grab the file from the .cab with something like this: mkdir foo cd foo cabextract ../vcredist_x64.exe cabextract vc_red.cab cp F_CENTRAL_mfc100_x86 ~/.wine/windows/system32 cd .. rm -rf foo That gets the app I'm trying to the point where it needs a newer builtin vcrun100, http://source.winehq.org/git/wine.git/?a=commit;h=6cb3713e26aa4d89d4dc67601… which is great. (Now I just have to get around to following up on previous posts about how to build biarch wine on ubuntu 12.04.) But I still wonder why the installer doesn't want to install the 64 bit files.

1 1

Re: comctl32: When created with TCS_VERTICAL, TCS_MULTILINE is set automatically
by Stefan Leichter May 17, 2012

May 17, 2012

Thursday 17 May 2012 Nikolay Sivov <nsivov(a)codeweavers.com> > - DWORD dwStyle; > + DWORD style; Any reason for changing the name?

2 1

Re: kernel32: check the last parameter of K32EnumProcessModules for NULL before use (with test)
by Nikolay Sivov May 17, 2012

May 17, 2012

On 5/17/2012 11:46, Stefan Leichter wrote: > SetLastError(0xdeadbeef); > + pEnumProcessModules(hpQI,&hMod, sizeof(HMODULE), NULL); > + ok(GetLastError() == ERROR_ACCESS_DENIED, "expected error=ERROR_ACCESS_DENIED but got %d\n", GetLastError()); > + > + SetLastError(0xdeadbeef); > + ret = pEnumProcessModules(hpQV,&hMod, sizeof(HMODULE), NULL); > + ok(!ret, "succeeded\n"); > + ok(GetLastError() == ERROR_NOACCESS, "expected error=ERROR_NOACCESS but got %d\n", GetLastError()); > + Why don't you check return value from the first call?

2 1

Re: [appdb] Applications working flawlessly using patched wine should be rated Gold
by K.King May 17, 2012

May 17, 2012

<< A "runs well with patched wine" checkbox might be useful, but would be very confusing. Maybe if it forced the rating to be garbage that would be ok. >> If it was possible my preference would be to be able to see separate rating per vanilla and per non vanilla. The per vanilla should be first and more obvious. If I see a rating of Garbage infers even with patching it would not run? problem with "runs well with patched wine" is it's an on or off, whereas you could start wishing for an indication of how well it can be made to run? Keeping the steps, results, capabilities, not tested, etc for patched wine could/should maybe be kept quite separate to the quick and obvious vanilla out of box results section? K.

1 0

kernel32: Correct WideCharToMultiByte and MultiByteToWideChar error codes and conditions
by Alex Henrie May 16, 2012

May 16, 2012

Hello, This patch was marked committed yesterday--thank you for accepting it!--but it still isn't showing up in Git. Maybe it was accidentally skipped? It needs to be committed before I can submit the UTF-7 implementation I wrote that builds on it. -Alex

2 1

AppDB, ratings and native vs. builtin trouble
by Alexey Loukianov May 16, 2012

May 16, 2012

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Good day to all subscribers of this list, I've got a small decision trouble I have to deal with as an AppDB maintainer. I know that years ago there had been discussion on this topic, but I want to bring this once on a surface once again so we would get a fresh modern look an thus minor problem. As we all know, Wine isn't perfect, thus applications often require native DLLs to behave properly. AppDb treats such cases as being "Gold" rating. It's easy to distinguish between "Platinum" and "Gold" case most of the time, but sometimes there are cases where native dll overrides happen automagically as a result of app installed installing third-party lib redists himself. Good example is idSoftware's Rage game which require XAudio2 to functions - which is not implemented in Wine. User would have to install and use native lib for game to function, but most of the times it would be done automatically by Steam (its "auto-install required libs" feature would try to install DirectX redist at the first time user would try to launch the game). How should we treat situations like that? From user PoV it's "Platinum" - app is working out of the box. From real side of things - it is "Gold", as native dll override is required for app to function - Wine's stubbed xaudio2 implementation is obviously not enough for game to work. With the recent release of Diablo III we're are at the same spot: game installer auto-installs vcrun2008 redist as a part of installation process, so end-user experience is like this game "works out of a box". Problem is that D3 installed isn't working unless user patches Wine's AcceptEx implementation with Erichs patchset, so for vast majority of Wine's userbase installing D3 would be something like "copy already installed D3 from other PC" - and that would bring the problem of the game requiring native VC++ 2008 runtime on the surface. How should we treat/assign ratings for these cases? P.S. Aside from that, I want to once again bring up the discussion on extending AppDB so more detailed test reports would be possible. IMO it might be reasonable to add to AppDB test report form is an ability to specify whether the version used was "vanilla" or "patched with some out-of-tree patches" one. For most reporters sane default would be "vanilla", while at some circumstances most of the test reports would come for "patched" version: good examples are SW:KOTR, WoT, D3 and many other games that require out-of-tree patches to function best under Wine. Thanks for spending your time reading this, and thanks in advance for your answers. - -- Best regards, Alexey Loukianov mailto:mooroon2(a)mail.ru System Engineer, Mob.:+7(926)218-1320 *nix Specialist -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.16 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJPs34GAAoJEPB9BOdTkBUL5KMH+gOtjQv1ckZ4tZ9X+kysZ/+0 gM1uCSHvuUdHuY+aVlUs688NU76mJcUu+ORPSuMrdltm7BCn5DbAN3ITjMSf2Y7y KHvvDFbGcKkiDupCyQhdNHpEyfC3PFzWjuI7o8nG/2eZGum5zRwGdbgvDvSc2Qo1 md208/6WDAGNjN2943orolG5pqhKapDvjvpoKSavgW1ykHeKDiTvCiNKXvcRU2ox bRZomHY93DNV0RoLnExcp8HnA/DooWQdGfwF4uTp+wgmIbwbWM4iw3Zw5CuXim+6 VzetXUPU9iVr2h3xqBgmVq5a7rMri7mykDXACo7ex30ekWM9XQBqSgT0jxqIi14= =16iP -----END PGP SIGNATURE-----

3 6

Re: [1/2] wmvcore: Add stub for WMCreateSyncReader (try 2) (resend)
by Nikolay Sivov May 16, 2012

May 16, 2012

On 5/16/2012 19:46, André Hentschel wrote: > + wmsdkidl.idl \ > + wmsbuffer.idl \ This is supposed to be sorted. Also I think you need to add .gitignore changes as well.

1 0

RE: localspl: Only store the file part of the filenames in the registry
by Detlef Riekenberg May 16, 2012

May 16, 2012

Hi Huw. We have only filenames in the registry. The current code works fine with the adobe postscript driver and various other driver for PDF printers. Which native driver makes problems? Why do you add spaces before and after the the function args and change idention, bracket style and variable names everywhere in the patch? HP is using a subdirectory for some driver files, but i can't remember the driver name at the moment. -- By by ... Detlef

2 1

re: AppDB, ratings and native vs. builtin trouble
by Dan Kegel May 16, 2012

May 16, 2012

16.05.2012 17:34, Rosanne DiMesio wrote: > If the user didn't have to manually do anything, it should be rated > platinum. Agreed. appdb ratings are from the point of view of someone who simply uses their linux distro to install wine, and then tries to install a game. If no fiddling is required, and the game works perfectly, it's platinum even if under the hood the game has snuck in a native (and Microsoft-copyrighted) DLL. (DLLs not copyrighted by Microsoft are not generally a concern.) In parallel, people who care about making Wine run without any native DLLs should make sure there's a "purist" bug in wine's bugzilla describing what breaks if you force wine to use its own builtin DLL for the game. (And should then try to fix that bug. :-) - Dan

2 1

Re: [2/2] gdiplus: Implement GdipImageGetFrameCount and GdipImageSelectActiveFrame.
by Vincent Povirk May 16, 2012

May 16, 2012

Er, you removed the FIXME, but GdipImageSelectActiveFrame is still a stub? You're not changing the image data.

2 1

Re: [PATCH 2/2] ieframe: Added GetExternal tests
by Marvin May 16, 2012

May 16, 2012

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=18421 Your paranoid android. === WNT4WSSP6 (32 bit webbrowser) === webbrowser.c:3280: Test failed: Creating WebBrowser object failed: 80040154 webbrowser.c:3280: Test failed: Creating WebBrowser object failed: 80040154 webbrowser.c:3280: Test failed: Creating WebBrowser object failed: 80040154 webbrowser.c:3280: Test failed: Creating WebBrowser object failed: 80040154 webbrowser.c:3402: Test failed: Creating WebBrowser object failed: 80040154 webbrowser.c:3377: Test failed: Could not get WebBrowserV1 instance: 80040154 webbrowser: unhandled exception c0000005 at 0040AE8E === W2KPROSP4 (32 bit webbrowser) === webbrowser.c:2594: Test failed: expected GetOverridesKeyPath webbrowser.c:2599: Test failed: expected Invoke_SETSECURELOCKICON webbrowser.c:2600: Test failed: expected Invoke_FILEDOWNLOAD webbrowser.c:3019: Test failed: doc_disp == NULL webbrowser: unhandled exception c0000005 at 00402FCA

1 0

Re: mstask: Actually run a test and fix the expected behaviour.
by Marvin May 16, 2012

May 16, 2012

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=18419 Your paranoid android. === WNT4WSSP6 (32 bit task_trigger) === task_trigger.c:165: Test failed: Failed to setup test_task

2 1

Re: [1/2] comctl32/tests: Added tests for treeview right mouse click handling
by Nikolay Sivov May 16, 2012

May 16, 2012

On 5/15/2012 22:53, Daniel Jelinski wrote: > > > + PostMessageA(hTree, WM_RBUTTONUP, 0, (LPARAM)0x10001); > + /* this sequence should NOT send WM_CONTEXTMENU */ > + SendMessageA(hTree, WM_RBUTTONDOWN, 2, (LPARAM)0x10001); Please use defined names for wParam and a MAKELPARAM macro to make a coordinate dword. > + test_mouse(); Also this deserves a more descriptive name, something like test_rbutton().

1 0

Re: mstask: Actually run a test and fix the expected behaviour.
by Marvin May 16, 2012

May 16, 2012

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=18419 Your paranoid android. === WNT4WSSP6 (32 bit task_trigger) === task_trigger.c:165: Test failed: Failed to setup test_task

1 0

Re: mstask/tests: Add missing call to the tested function.
by Marvin May 16, 2012

May 16, 2012

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=18417 Your paranoid android. === WNT4WSSP6 (32 bit task_trigger) === task_trigger.c:165: Test failed: Failed to setup test_task === W2KPROSP4 (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === WXPPROSP3 (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === W2K3R2SESP2 (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === WVISTAADM (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === W2K8SE (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === W7PRO (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === W7PROX64 (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === TEST64_W7SP1 (32 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === W7PROX64 (64 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000 === TEST64_W7SP1 (64 bit task_trigger) === task_trigger.c:307: Test failed: Expected E_INVALIDARG: 0x00000000

1 0