December 2009 - Wine-Devel - list.winehq.org

Re: ddraw: Fix an incorrect refcount test
by Stefan Dösinger Dec. 17, 2009

Dec. 17, 2009

Am 17.12.2009 um 20:20 schrieb Luke Benstead: > Luke. > <0001-ddraw-Fix-an-incorrect-refcount-test.txt> This patch is correct. The test intends to test DDraw7, there's a DDraw4 test below. So the variable name was wrong, not the macro.

1 0

Re: [shell32/tests] Initialize an APPBARDATA struct (Valgrind)
by Alexandre Julliard Dec. 17, 2009

Dec. 17, 2009

Paul Vriens <paul.vriens.wine(a)gmail.com> writes: > Fixes some Valgrind warnings. > > Changelog > Initialize an APPBARDATA struct (Valgrind) That's just hiding the problem, an app can pass the same structure. -- Alexandre Julliard julliard(a)winehq.org

2 3

Re: [usrmarshal] Adjust new tests to pass on win9x platforms.
by Jacek Caban Dec. 17, 2009

Dec. 17, 2009

Hi Jeremy, On 12/17/09 12:13 AM, Jeremy White wrote: > --- > dlls/oleaut32/tests/usrmarshal.c | 9 ++++++--- > 1 files changed, 6 insertions(+), 3 deletions(-) > > diff --git a/dlls/oleaut32/tests/usrmarshal.c b/dlls/oleaut32/tests/usrmarshal.c > index e9c53e9..1c770de 100644 > --- a/dlls/oleaut32/tests/usrmarshal.c > +++ b/dlls/oleaut32/tests/usrmarshal.c > @@ -306,7 +306,8 @@ static void test_marshal_LPSAFEARRAY(void) > buffer = HeapAlloc(GetProcessHeap(), 0, size); > init_user_marshal_cb(&umcb,&stub_msg,&rpc_msg, buffer, size, MSHCTX_DIFFERENTMACHINE); > next = LPSAFEARRAY_UserMarshal(&umcb.Flags, buffer,&lpsa); > - ok(next - buffer == expected, "Marshaled %u bytes, expected %u\n", (ULONG) (next - buffer), expected); > + ok(next - buffer == expected || next - buffer + sizeof(DWORD) == expected /*win9x*/, > + "Marshaled %u bytes, expected %u\n", (ULONG) (next - buffer), expected); > It looks like a good place to use broken(). Thanks, Jacek

3 2

Re: [PATCH 6/7] urlmon: Report redirected URL to object created by BindToObject.
by Paul Vriens Dec. 17, 2009

Dec. 17, 2009

On 12/15/2009 09:16 PM, Jacek Caban wrote: > --- > dlls/urlmon/binding.c | 6 +++++ > dlls/urlmon/tests/url.c | 58 ++++++++++++++++++++++++++++++---------------- > 2 files changed, 44 insertions(+), 20 deletions(-) > Hi Jacek, This one introduced a test failure: http://test.winehq.org/data/tests/urlmon:url.html I didn't find anything common for the boxes/OS, race? Could you have a look? -- Cheers, Paul.

4 4

Re: [PATCH] secur32: Fix memory leak in SECUR32_addPackages. (valgrind)
by James Hawkins Dec. 17, 2009

Dec. 17, 2009

On Wed, Dec 16, 2009 at 5:07 PM, Lei Zhang <thestig(a)google.com> wrote: > Hi, this is for bug 21045. > + SecurePackage *prev_package = NULL; LIST_FOR_EACH_ENTRY(package, &packageTable->table, SecurePackage, entry) { + if (prev_package) + HeapFree(GetProcessHeap(), 0, prev_package); + prev_package = package; HeapFree(GetProcessHeap(), 0, package->infoW.Name); HeapFree(GetProcessHeap(), 0, package->infoW.Comment); } + if (prev_package) + HeapFree(GetProcessHeap(), 0, prev_package); I believe the construct you're looking for is LIST_FOR_EACH_SAFE. grep through the code for examples on using that to safely free and remove elements from a list. -- James Hawkins

2 1

Re: [1/2] gdiplus: Explicitly copy the bits from dib sections to new bitmaps.
by Paul Vriens Dec. 17, 2009

Dec. 17, 2009

On 12/13/2009 12:59 AM, Vincent Povirk wrote: > Hi Vincent, It appears that this one introduced a test failure: http://test.winehq.org/data/tests/gdiplus:image.html One of the tests that fails now (line 533) was there for a long time it's just that you've added some stuff before that. The other one (line 1058) is because of the other patch "Add test for image palette functions." Could you have a look? -- Cheers, Paul.

2 5

Re: [PATCH] inetcomm/smtptransport.c : Remove some unneeded variables and assigns.
by Juan Lang Dec. 16, 2009

Dec. 16, 2009

Hi Amine, this patch has no functional benefit. For example, - HRESULT hr; TRACE("\n"); - hr = SMTPTransport_ParseResponse(This, pBuffer, &response); - if (FAILED(hr)) + if (FAILED(SMTPTransport_ParseResponse(This, pBuffer, &response))) This has the dubious benefit of removing a local variable that was used before. I can't fathom why this is a good thing. --Juan

4 7

Re: [PATCH] inetcomm/smtptransport.c : Fix a typo ?
by Juan Lang Dec. 16, 2009

Dec. 16, 2009

Hi Amine, -static void SMTPTransport_CallbackMessageSendTo(IInternetTransport *iface, char *pBuffer, int cbBuffer); - > The actual function is a little bit under. Um, it's called a prototype, and it's needed to compile smtptransport.c. Have you tested this patch at all? --Juan

2 2

Re: inetcomm/smtptransport.c : Remove some unneeded variables and assigns.
by Austin English Dec. 16, 2009

Dec. 16, 2009

On Wed, Dec 16, 2009 at 12:28 PM, Amine Khaldi <amine48rz(a)gmail.com> wrote: > WBR, > Amine Khaldi. Out of curiosity, what tool are you using to find these bugs? -- -Austin

3 3

Re: hlink/hlink_main.c : Remove an unneeded assignment.
by Jacek Caban Dec. 16, 2009

Dec. 16, 2009

Hi Amine, > r = register_clsid(&CLSID_StdHlink); > if (SUCCEEDED(r)) > - r = register_clsid(&CLSID_StdHlinkBrowseContext); > + register_clsid(&CLSID_StdHlinkBrowseContext); > > return S_OK; In this case the correct fix is to return r. Please be more careful then doing janitorial work. There are more places in your patches that seem questionable. You should make sure that we want to ignore an error instead of handling it (and usually we don't) before deleting such assignments. Otherwise it's just hiding bugs. Thanks, Jacek

2 1

Re: inetcomm/smtptransport.c : Remove some unneeded variables and assigns.
by Austin English Dec. 16, 2009

Dec. 16, 2009

On Wed, Dec 16, 2009 at 12:28 PM, Amine Khaldi <amine48rz(a)gmail.com> wrote: > WBR, > Amine Khaldi. Out of curiosity, what tool are you using to find these bugs? -- -Austin

1 0

Re: [PATCH 07/10] urlmon: Added PARSE_DOMAIN to CoInternetParseUrl
by Paul Vriens Dec. 16, 2009

Dec. 16, 2009

On 12/15/2009 11:47 PM, Piotr Caban wrote: > --- > dlls/urlmon/internet.c | 32 ++++++++++++++++++++++++++++++++ > dlls/urlmon/tests/misc.c | 23 +++++++++++++++++------ > 2 files changed, 49 insertions(+), 6 deletions(-) > > Hi Piotr, This one has introduced some test failures: http://test.winehq.org/data/tests/urlmon:misc.html Looks more OS than IE specific? Could you have a look? -- Cheers, Paul.

1 0

chromium wine valgrind bot up...
by Dan Kegel Dec. 16, 2009

Dec. 16, 2009

More progress on my http://wiki.winehq.org/UnitTestSuites effort: Lei and I are running a buildbot that runs win32 chromium's unit tests under wine under valgrind. It just came up yesterday, and we're still working out some kinks, but you can see it at http://build.chromium.org/buildbot/waterfall.fyi/builders/Chromium%20Linux%… Running a real app's test suite under valgrind+wine can find both bugs in wine and bugs in the app; here are examples of both found today by that bot: http://bugs.winehq.org/show_bug.cgi?id=21041 (mismatched GlobalLock/HeapFree in wine?) http://code.google.com/p/chromium/issues/detail?id=30584 (app forgot to initialize a bitmap?) It'll be interesting to see what else this turns up over time.

1 0

Re: [try 2] [2/7] mapi32: Correct the Simple MAPI parameters
by Henri Verbeet Dec. 16, 2009

Dec. 16, 2009

2009/12/16 Owen Rudge <orudge(a)codeweavers.com>: > +214 stdcall MAPIFindNext(ptr ptr str str long long str) The last parameter is an output parameter.

1 0

Re: [2/5] winemapi: Add Simple MAPI functions
by Henri Verbeet Dec. 16, 2009

Dec. 16, 2009

2009/12/15 Owen Rudge <orudge(a)codeweavers.com>: > + 1 stdcall MAPIAddress(ptr ptr ptr long ptr long long ptr long ptr ptr) > + 6 stdcall MAPILogoff(long long long long) > +11 stdcall MAPISendMail(long long ptr long long) These don't look quite right to me. Somewhat related, LHANDLE is a ULONG in include/mapi.h, but a ULONG_PTR in my copy of the PSDK.

2 1

Re: msvcrtd: free mem after use (valgrind)
by Paul Vriens Dec. 16, 2009

Dec. 16, 2009

On 12/16/2009 03:25 PM, André Hentschel wrote: > --- > dlls/msvcrtd/tests/debug.c | 2 ++ > 1 files changed, 2 insertions(+), 0 deletions(-) > > diff --git a/dlls/msvcrtd/tests/debug.c b/dlls/msvcrtd/tests/debug.c > index 71f6e4c..dc89230 100644 > --- a/dlls/msvcrtd/tests/debug.c > +++ b/dlls/msvcrtd/tests/debug.c > @@ -60,6 +60,8 @@ static void test_new(void) > > mem = pMSVCRTD_operator_new_dbg(42, _NORMAL_BLOCK, __FILE__, __LINE__); > ok(mem != NULL, "memory not allocated\n"); > + if(mem != NULL) > + HeapFree(GetProcessHeap(), 0, mem); > } Hi André, There is no need to check before calling HeapFree. -- Cheers, Paul.

2 1

Re: msvcrtd: free mem after use (valgrind)
by Nikolay Sivov Dec. 16, 2009

Dec. 16, 2009

On 12/16/2009 17:25, André Hentschel wrote: > --- > dlls/msvcrtd/tests/debug.c | 2 ++ > 1 files changed, 2 insertions(+), 0 deletions(-) > > diff --git a/dlls/msvcrtd/tests/debug.c b/dlls/msvcrtd/tests/debug.c > index 71f6e4c..dc89230 100644 > --- a/dlls/msvcrtd/tests/debug.c > +++ b/dlls/msvcrtd/tests/debug.c > @@ -60,6 +60,8 @@ static void test_new(void) > > mem = pMSVCRTD_operator_new_dbg(42, _NORMAL_BLOCK, __FILE__, __LINE__); > ok(mem != NULL, "memory not allocated\n"); > + if(mem != NULL) > + HeapFree(GetProcessHeap(), 0, mem); > } > > /**********************************************************************/ > Pointer check is redundant in this case.

1 0

Re: [PATCH 2/3] mmdevapi: Add stubs for MMDevEnum, with tests
by Alexandre Julliard Dec. 16, 2009

Dec. 16, 2009

Maarten Lankhorst <m.b.lankhorst(a)gmail.com> writes: > +static ULONG WINAPI MMDevEnum_Release(IMMDeviceEnumerator *iface) > +{ > + MMDevEnumImpl *This = (MMDevEnumImpl*)iface; > + LONG ref = InterlockedDecrement(&This->ref); > + TRACE("Refcount now %i\n", ref); > + return ref; > +} Sounds like you want a static object here too... -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [3/5] WineD3D: Prepare for dynamic vertex buffers
by Henri Verbeet Dec. 16, 2009

Dec. 16, 2009

2009/12/15 Stefan Dösinger <stefan(a)codeweavers.com>: > + /* buffer_get_sysmem can remove the BO from dynamic buffers */ No, it can't? > @@ -540,8 +547,8 @@ static HRESULT WINAPI IWineD3DDeviceImpl_CreateVertexBuffer(IWineD3DDevice *ifac > TRACE("Not creating a vbo because GL_ARB_vertex_buffer is not supported\n"); > } else if(Pool == WINED3DPOOL_SYSTEMMEM) { > TRACE("Not creating a vbo because the vertex buffer is in system memory\n"); > - } else if(Usage & WINED3DUSAGE_DYNAMIC) { > - TRACE("Not creating a vbo because the buffer has dynamic usage\n"); > + } else if(!gl_support_dynamic_buffer(gl_info) && (Usage & WINED3DUSAGE_DYNAMIC)) { > + TRACE("Not creating a vbo because the buffer has dynamic usage but GL doesn't support dynamic buffers\n"); > } else if(!(Usage & WINED3DUSAGE_OPTIMIZE) && conv) { > TRACE("Not creating a vbo because the fvf needs conversion, but VB optimization is disabled\n"); > } else { It seems to me that this and similar code in CreateIndexBuffer() should be in buffer_init() instead.

2 1

Edit control messages through Comctl32 v6 module
by Nikolay Sivov Dec. 16, 2009

Dec. 16, 2009

Hi. Yesterday I looked at XP introduced messages for Edit controls (EM_SETCUEBANNER/EM_GETCUEBANNER). This definitely doesn't work without v6 module loaded. Does someone have any idea how this supposed to work? Any comments are welcome as usual. P.S. Yes, I know that comctl32 does subclass user32 control but only if theming is active (I mean Wine behavior), anyway this subclass procedure only cares about painting messages, control specific messages require control data manipulation, and this isn't a preferable of course.

4 18

Re: [PATCH 10/10] urlmon: Added tests for CoInternetGetSecurityUrl
by Paul Vriens Dec. 16, 2009

Dec. 16, 2009

On 12/15/2009 11:48 PM, Piotr Caban wrote: > --- > dlls/urlmon/tests/sec_mgr.c | 31 +++++++++++++++++++++++++++++++ > 1 files changed, 31 insertions(+), 0 deletions(-) > > Hi Piotr, Is there a way to avoid using lstrmcpW as that will always fail (or the !lstrcmpW case will always succeed) on Win9x/WinMe? It looks like the rest of this file use memcmp to compare strings. -- Cheers, Paul.

2 2

Re: [10/10] rpcrt4: Add tests for RPC calls with authentication.
by Paul Vriens Dec. 16, 2009

Dec. 16, 2009

Hi Rob, I took the liberty of running these tests on winetestbot. It shows two errors for W2K and above: https://winetestbot.geldorp.nl/JobDetails.pl?Key=115 -- Cheers, Paul.

3 7

strangeness under Dec 14th Wine Git
by chris ahrendt Dec. 15, 2009

Dec. 15, 2009

Just a little nit I noticed with yesterdays Git tree... When you give a window other than the wine window the focus... the wine window does not go into the background... so IF I have 2 or 3 windows open with a wine app running and click on say thunderbird... or firefox to look something up I have to move the wine window to the bottom of the screen. It looses focus (turns grey) but does not go to the background... Chris

1 0

[rfc] Provide separate operator vector new and delete in msvcrt; let valgrind check them
by Dan Kegel Dec. 15, 2009

Dec. 15, 2009

First draft for discussion. Not, um, well tested yet. ANSI C++ mandates separate vector versions of new and delete that do the same thing as the non-vector versions, but which can be replaced by user apps (see http://www.csci.csusb.edu/dick/c++std/cd2/lib-support.html#lib.new.delete ) So far, so good; wine does that in msvcrt. However, once in a while, a programmer will call delete when he should have called delete[], and a bunch of destructors won't get called. To detect this programming error, let's split the vector new and delete functions off from their scalar twins, and add values for the heap function flags so we can inform valgrind what kind of blocks are being wrangled. This means picking two flag bits that Microsoft doesn't use; there seem to be enough to go around. Requires the corresponding patch to valgrind ttps://bugs.kde.org/show_bug.cgi?id=218741 Thanks to Steve Vandebogart for his help on this; he wrote a bunch of it.

2 1

Tests for GetLongPathName
by Paul Vriens Dec. 15, 2009

Dec. 15, 2009

Hi Alexandre, Several file related functions have a maximum path size of 32767 when using a "\\?\" prefix. Testing these large paths only works on Windows due to Linux having a maximum path limit of 4096. Does it still make sense to add these tests or should I focus on <= 4096? I guess if we/Wine have a limit due to the OS we are running on we should probably act the same as that higher limit on Windows? -- Cheers, Paul.

2 1

Re: [PATCH 1/3] mmdevapi: Add class factory
by Alexandre Julliard Dec. 15, 2009

Dec. 15, 2009

Maarten Lankhorst <m.b.lankhorst(a)gmail.com> writes: > +static ULONG WINAPI MMCF_Release(LPCLASSFACTORY iface) > +{ > + IClassFactoryImpl *This = (IClassFactoryImpl *)iface; > + ULONG ref = InterlockedDecrement(&(This->ref)); > + TRACE("(%p) ref was %d\n", This, ref + 1); > + /* static class, won't be freed */ You don't need a refcount then. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: gdiplus: Fix a memory leak in the tests.
by Nikolay Sivov Dec. 15, 2009

Dec. 15, 2009

On 12/15/2009 14:35, Huw Davies wrote: > Found by Valgrind. > --- > dlls/gdiplus/tests/customlinecap.c | 1 + > 1 files changed, 1 insertions(+), 0 deletions(-) Hi, Huw. This leak will be fixed with corrected GdipCreateCustomLineCap(). Return code should never be Ok here. P.S. Same with previous patch for GdipCreateFontFamilyFromNam().

2 1

Re: [3/3] WineD3D: Don't enable render_to_fbo on single buffered swapchains
by Henri Verbeet Dec. 15, 2009

Dec. 15, 2009

2009/12/15 Stefan Dösinger <stefan(a)codeweavers.com>: > > Am 15.12.2009 um 08:14 schrieb Henri Verbeet: > >> 2009/12/15 Stefan Dösinger <stefan(a)codeweavers.com>: >>> + && present_parameters->BackBufferCount > 0 >> "BackBufferCount" is unsigned. > Not even -Wall -pedantic complains about this, but okay. This patch compares against unsigned constants. > My point was just that you might as well just write "present_parameters->BackBufferCount", since it's never going to be below zero.

1 0

Re: msi: Replace the string table hash with a sorted index.
by Hans Leidekker Dec. 15, 2009

Dec. 15, 2009

On Tuesday 15 December 2009 09:51:20 Hans Leidekker wrote: > Addresses a performance bottleneck in the Visual Studio 2005 installer. > See http://bugs.winehq.org/show_bug.cgi?id=14168 > --- > dlls/msi/msipriv.h | 2 +- > dlls/msi/string.c | 147 ++++++++++++++++++++++++++++++++++++++-------------- > 2 files changed, 109 insertions(+), 40 deletions(-) Please disregard this patch, Henri pointed out some improvements that can be made. -Hans

1 0

winetricks directx testers needed
by Austin English Dec. 15, 2009

Dec. 15, 2009

Howdy all, I e-mailed wine-users a few days ago, but had no volunteers. I know most of you here aren't the target audience for using winetricks directx9, but since neither Dan or I game much, we need a guinea pig ;-). Long story short, someone complained that 'winetricks directx9' crashes on OS X, but works if the windows version is set to xp instead of 2k. I've played around with it a bit, and it seems that using xp gives a different set of installed files. Which one works better for games, however, I don't know. So, test it please. If you get wine from SVN, there's a beta of it: http://winezeug.googlecode.com/svn/trunk/winetricks just use 'winetricks directx9-beta' for the XP version, and 'winetricks directx9' for windows 2k (on OS X, it forces directx9-beta). Please let me know which version works better for games that need it. Any problems, file a bug at winezeug (http://code.google.com/p/winezeug/issues/list) For faster service, shoot me an e-mail, since I don't get e-mail notifications of any new bugs there... -- -Austin

3 6

Re: [PATCH 2/5] wined3d: Filter window messages generated by switching to fullscreen and back.
by Stefan Dösinger Dec. 15, 2009

Dec. 15, 2009

This looks correct to me, I'm just asking to make sure I understand this correctly: > + filter_messages = swapchain->filter_messages; > + swapchain->filter_messages = TRUE; > + > SetWindowLongW(window, GWL_STYLE, style); > SetWindowLongW(window, GWL_EXSTYLE, exstyle); > SetWindowPos(window, HWND_TOP, 0, 0, w, h, SWP_FRAMECHANGED | SWP_SHOWWINDOW); > + > + swapchain->filter_messages = filter_messages; > } I am not seeing any place where swapchain->filter_messages is set so something other than FALSE, except of the same code in restore_fullscreen_window. Am I right to assume that this construct is meant to catch situations where the app's wndproc calls wined3d when receiving a WM_DISPLAYCHANGE message? > @@ -965,6 +976,13 @@ void get_drawable_size_swapchain(struct wined3d_context *context, UINT *width, U > LRESULT swapchain_process_message(IWineD3DSwapChainImpl *swapchain, HWND window, > UINT message, WPARAM wparam, LPARAM lparam, WNDPROC proc) > { > + if (swapchain->filter_messages) > + { > + TRACE("Filtering message: window %p, message %#x, wparam %#lx, lparam %#lx.\n", > + window, message, wparam, lparam); > + return CallWindowProcW(DefWindowProcW, window, message, wparam, lparam); > + } > + This calls the default Win32 wndproc in user32, and thus bypasses the app's wndproc, correct? There's the other call below added by patch 1 which calls proc instead. I'm asking because the lack of a ! in the if condition confused me at first.

2 1

Re: [2/3] WineD3D: Properly release client storage memory
by Henri Verbeet Dec. 15, 2009

Dec. 15, 2009

2009/12/15 Stefan Dösinger <stefan(a)codeweavers.com>: > + This->Flags &= ~(SFLAG_ALLOCATED | SFLAG_SRGBALLOCATED); That's pretty much what surface_force_reload() does.

1 0

Re: [3/3] WineD3D: Don't enable render_to_fbo on single buffered swapchains
by Henri Verbeet Dec. 15, 2009

Dec. 15, 2009

2009/12/15 Stefan Dösinger <stefan(a)codeweavers.com>: > + && present_parameters->BackBufferCount > 0 "BackBufferCount" is unsigned.

1 0

Test suite under Intel drivers
by Jaime Rave Dec. 14, 2009

Dec. 14, 2009

Hi wine devs, I'm writing you to ask some help to try to fix the the Wine test suite under the Intel drivers. I have already fill a bug with freedesktop but they are asking for some wine devs to give them some openGL test to add it to the piglit testcases. The bug is: http://bugs.freedesktop.org/show_bug.cgi?id=25432 Some words from the bug: "There are probably some actual bugs in our OpenGL, and there may also be bugs in the testcase. What it'll take to fix it is somebody (probably Wine developers/fans) isolating some OpenGL testcases for inclusion in piglit that will catch the bugs that this D3D test is hitting in OpenGL." Thanks in advance for your help. -- Jaime Rave Torres Ingeniero de Sistemas Miembro de la Fundación de Software Libre de Colombia http://www.fslcol.org/ http://sobrelibertad.blogspot.com/ 3003160361

2 1

Re: [PATCH 2/2] mmdevapi: Add more tests
by Paul Vriens Dec. 14, 2009

Dec. 14, 2009

On 12/14/2009 10:09 AM, Maarten Lankhorst wrote: > + hr = IMMDeviceEnumerator_GetDefaultAudioEndpoint(mme, eRender, eMultimedia,&dev); > + ok(hr == S_OK, "GetDefaultAudioEndpoint failed: 0x%08x\n", hr); > + if (hr != S_OK) > + { > + skip("GetDefaultAudioEndpoint returns 0x%08x, skipping tests\n", hr); > + return; > + } Hi Maarten, This one introduces a failure on Vista+ (obviously) for boxes with a soundcard so it seems (by looking at other tests for these boxes): http://test.winehq.org/data/tests/mmdevapi:dependency.html Could you have a look? -- Cheers, Paul.

2 1

Re: [PATCH 14/14] [WineDbg]: fix some casts issues on 64bit platform
by Alexandre Julliard Dec. 14, 2009

Dec. 14, 2009

Eric Pouech <eric.pouech(a)orange.fr> writes: > diff --git a/programs/winedbg/dbg.y b/programs/winedbg/dbg.y > index 855407f..7ed6307 100644 > --- a/programs/winedbg/dbg.y > +++ b/programs/winedbg/dbg.y > @@ -281,9 +281,9 @@ info_command: > | tINFO tCLASS { info_win32_class(NULL, NULL); } > | tINFO tCLASS tSTRING { info_win32_class(NULL, $3); } > | tINFO tWND { info_win32_window(NULL, FALSE); } > - | tINFO tWND expr_rvalue { info_win32_window((HWND)$3, FALSE); } > + | tINFO tWND expr_rvalue { info_win32_window((HWND)(DWORD_PTR)$3, FALSE); } > | tINFO '*' tWND { info_win32_window(NULL, TRUE); } > - | tINFO '*' tWND expr_rvalue { info_win32_window((HWND)$4, TRUE); } > + | tINFO '*' tWND expr_rvalue { info_win32_window((HWND)(DWORD_PTR)$4, TRUE); } > | tINFO tPROCESS { info_win32_processes(); } > | tINFO tTHREAD { info_win32_threads(); } > | tINFO tEXCEPTION { info_win32_exceptions(dbg_curr_tid); } It seems to me that it's expr_rvalue itself that should be a DWORD_PTR, it can be used to specify addresses at other places too. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: [usrmarshal] Check the actual size of safe arrays marshaled, not just projected size.
by Paul Vriens Dec. 14, 2009

Dec. 14, 2009

On 12/11/2009 04:25 PM, Jeremy White wrote: > --- > dlls/oleaut32/tests/usrmarshal.c | 41 +++++++++++++++++++++++-------------- > 1 files changed, 25 insertions(+), 16 deletions(-) > Hi Jeremy, This one looks like to have introduced some test failures on Win9x/WinME: http://test.winehq.org/data/tests/oleaut32:usrmarshal.html Could you have a look? -- Cheers, Paul.

1 0

RE: WineTestBot
by Greg Geldorp Dec. 14, 2009

Dec. 14, 2009

Hi Andre, > From: Andre Hentschel > > Some Questions/Suggestions by me: > - Are the VMs running on Computer at VMWare, on some Internetserver or at your home? I have a personal box (Dell SC1435, Quad core AMD Opteron @ 2.2 GHz, 16GB RAM) collocated at a large Dutch ISP. WineTestBot and the test VMs are all running on that box. VMware is not involved (yet?). > - IMO you should add pruning to your script, we are at about 120 tests and i guess old ones can be deleted Yes, I started working on a Janitor.pl script last weekend. Should be ready some time this week. The plan is to throw everything away that's older than a week. > - Will you add one or more Wine-Machines to cover that, too? It's not a priority for me, I might look at that once I get the bot to phase 3 (where it will watch wine-patches for test changes and run them automatically). Thanks for your feedback, Ge.

1 0

Re: Rob Shearman : msvcrt: Free memory allocated in TLS slot on module unload as well as thread exit.
by Nikolay Sivov Dec. 14, 2009

Dec. 14, 2009

On 12/14/2009 18:51, Alexandre Julliard wrote: > Module: wine > Branch: master > Commit: c20868e0a2f7af909cf8af2877ae8b024fa6d11e > URL: http://source.winehq.org/git/wine.git/?a=commit;h=c20868e0a2f7af909cf8af287… > > Author: Rob Shearman<robertshearman(a)gmail.com> > Date: Mon Dec 14 14:13:57 2009 +0000 > > msvcrt: Free memory allocated in TLS slot on module unload as well as thread exit. > > --- > > dlls/msvcrt/main.c | 27 +++++++++++++++------------ > 1 files changed, 15 insertions(+), 12 deletions(-) > It's a partial solution I suppose? I works for a one thread only that unloads module.

2 2

WineTestBot
by André Hentschel Dec. 14, 2009

Dec. 14, 2009

Hi, First i have to thank Mr. Geldorp for this really great gift :) Some Questions/Suggestions by me: - Are the VMs running on Computer at VMWare, on some Internetserver or at your home? - IMO you should add pruning to your script, we are at about 120 tests and i guess old ones can be deleted - Will you add one or more Wine-Machines to cover that, too? -- Best Regards, André Hentschel

1 0

Re: meaning of "Hardware" field in bugzilla?
by Dan Kegel Dec. 14, 2009

Dec. 14, 2009

Sure, once it's far enough along... might be a bit early yet? On Dec 11, 2009 6:15 PM, "Tom Wickline" <twickline(a)gmail.com> wrote: On Sat, Dec 12, 2009 at 2:35 AM, Austin English <austinenglish(a)gmail.com> wrote: > > > Dan and I dis... If your going to re-work this I would also add ARM to the list, their has been some work to port Wine to the ARM processor lately. Anyone have any objections to adding a ARM selection to the list? ARM -> RISC Cheers, Tom

4 5

Re: msi: Add binary search to the string table.
by Henri Verbeet Dec. 14, 2009

Dec. 14, 2009

2009/12/14 Hans Leidekker <hans(a)codeweavers.com>: > Addresses a performance bottleneck in the Visual Studio 2005 installer. > See http://bugs.winehq.org/show_bug.cgi?id=14168 Does it really make sense to maintain both a hash table and a binary search tree?

3 2

re: ntdll: Ignore the hole in the FILE_PIPE_WAIT_FOR_BUFFER structure passed into NtFsControlFile before writing the structure to the server.
by Dan Kegel Dec. 14, 2009

Dec. 14, 2009

Rob's patch is the most precise solution. The macro should probably go in a .h file as it might be useful elsewhere. Alternately I can suppress all errors from functions like server_ioctl_file but it'd be better to just mark the holes as defined. Thanks, Rob!

1 0

weird and funky things happen when viewing guimark in ie6 in wine...
by Dan Kegel Dec. 14, 2009

Dec. 14, 2009

Austin has a workaround for the problem that makes ie6 exit immediately ( http://code.google.com/p/winezeug/issues/detail?id=78 ). While testing it on guimark, wine ~.wine/drive_c/Program\ Files/Internet\ Explorer/IEXPLORE.EXE http://www.craftymind.com/factory/guimark/GUIMark_HTML4.html I noticed something strange when I closed ie6 after running for a few seconds. Either I see some sort of file leak (and this is with a fully vanilla fresh wine): ... wine client error:10c: pipe: Too many open files fixme:wininet:InternetSetOptionW INTERNET_OPTION_COOKIES_3RD_PARTY; STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_REQUEST_PRIORITY (-1): STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CODEPAGE (65001): STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_ERROR_MASK(11): STUB wine client error:10d: pipe: Too many open files fixme:wininet:InternetSetOptionW INTERNET_OPTION_COOKIES_3RD_PARTY; STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_REQUEST_PRIORITY (-1): STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CODEPAGE (65001): STUB fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_ERROR_MASK(11): STUB ... or I see Building font metrics. This may take some time... Font metrics: 0.0% done ... and the whole long font metric rebuilding process, which should never be happening anymore for local X servers (right?). The "wine client error" comes from ntdll/server.c. It ought to be easy to get a backtrace from the problem using winedbg by setting a breakpoint, but I haven't managed to so far. Debugging this is complicated by the fact that GuiMark seems to crash my home router; maybe it does too many fetches too quickly. Do other people see this, too? Firefox seems to handle it well in wine, and doesn't show this problem. If this turns out to be an ie6 bug, we don't care, but there's some chance it's a wine bug. Guimark seems to be quite the stress test. - Dan

3 2

Re: (resend)shlwapi: implement SHPropertyBag_ReadLONG
by Jacek Caban Dec. 14, 2009

Dec. 14, 2009

Hi Aric, On 12/11/09 10:44 PM, Aric Stewart wrote: > --- > dlls/shlwapi/ordinal.c | 25 +++++++++++++++++++++++++ > dlls/shlwapi/shlwapi.spec | 2 +- > 2 files changed, 26 insertions(+), 1 deletions(-) + hr = IPropertyBag_Read(ppb, pszPropName,&var, NULL); + if (SUCCEEDED(hr)) + *pValue = V_I4(&var); You should make sure that var is of VT_I4 type and return an error or do conversion if it's not. Thanks, Jacek

2 1

Re: ws2_32: Fix allocating a hostent with no h_aliases list.
by Alexandre Julliard Dec. 14, 2009

Dec. 14, 2009

Erich Hoover <ehoover(a)mines.edu> writes: > @@ -4616,11 +4620,18 @@ static struct WS_hostent *WS_create_he(char *name, int aliases, int addresses, i > sizeof(char *)*aliases + > sizeof(char *)*addresses); > > - /* Place addresses in the allocated memory, making sure to have enough > - * room for the NULL at the end of the list. > + /* NOTE: Even if the alias and address lists are "empty" the list pointer must > + * be non-NULL and contain a single NULL item. > */ > + if (aliases == 0 || addresses == 0) > + { > + FIXME("Cannot create a hostent with an empty list pointer!\n"); > + return NULL; > + } That's an internal constraint in the code, it doesn't make sense to have a FIXME. If you really want a check it should be an assert(). -- Alexandre Julliard julliard(a)winehq.org

1 0

Upcoming WWN
by Zachary Goldberg Dec. 14, 2009

Dec. 14, 2009

Greetings Folks, As is the usual excuse "real life" has gotten in the way of producing quality and timely World Wine News issues. Dan Kegel recently had the idea of putting together a wiki page where the community as a whole can contribute to the next issue. I concurred and Dan put together the following page: http://wiki.winehq.org/RecentNewsQueue Which has a preface as follows: "This is the inbox for Wine Weekly News. The WWN editor invites Wine developers and users to submit articles here for inclusion in the next edition of WWN, but reserves the right to edit mercilessly. This page will be cleared each time a new WWN is published.' So, feel free to throw up some topics you would like to see discussed and maybe some extra detail that wouldn't ordinarily have showed up on wine-devel (you can include links to wine-devel too if you feel like being really nice to the editor :)). I'll post some ideas for articles during the day tomorrow (something about dsound and OpenAL anybody?) that hopefully we can flesh out along with ideas from the community at large. Cheers, -Zach

1 0

New winetricks 20091213: ie6 and gecko fixed; new verbs vcrun6sp6, python-comtypes, richtx32; less restrictive art2kmin, clearer license text.
by Dan Kegel Dec. 14, 2009

Dec. 14, 2009

Austin and Andrew have been hard at work keeping winetricks working smoothly. In particular, the ie6 and gecko verbs should work again. Thanks guys! Changes since 20091125: Austin English: Workaround wine bug for IE6 Update shockwave sha1sum Add python-comtypes verb Explain windows licensing a bit better Gdiplus no longer comes from powerpoint viewer, update description accordingly Update DivX to latest version, remove obsolete /silent option Make append_path work if called more than once When installing gecko, regsvr32 mshtml afterward, so it actually works Update KDE to 0.9.6-5 When using native mdac, also override oledb32 Bump artk2min to 2007 Access runtime (freely redistributable, wooo!) Andrew Nguyen: Update flash to latest version. Check more exit codes. Update the vb6run verb to use the Service Pack 6 variant. Revise vcrun6 verb to remove more fake DLLs and more simply extract mfc42u.dll. Add vcrun6sp6 verb. (New verb rather than update of vcrun6 as it's a huge download.) Add richtx32 verb. Dan Kegel: Add heapcheck verb (for use with the not-yet-committed heapcheck patches) Avoid double-trying regedit Online as always at http://kegel.com/wine/winetricks or http://winezeug.googlecode.com (Bug reports to the issue tracker at the above URL, please.)

1 0

Re: mshtml: Implement IHTMLImgElement get_Name
by Jacek Caban Dec. 13, 2009

Dec. 13, 2009

Hi Alistair, On 12/13/09 12:11 PM, Alistair Leslie-Hughes wrote: > Hi, > > > Changelog: > mshtml: Implement IHTMLImgElement get_Name + nsres = nsIDOMHTMLImageElement_GetName(This->nsimg,&src_str); + if(NS_FAILED(nsres)) { + ERR("GetName failed: %08x\n", nsres); + return E_FAIL; + } You leak src_str here. +#define test_img_name(u) _test_img_name(__LINE__,u) +static void _test_img_name(unsigned line, IUnknown *unk) +{ + IHTMLImgElement *img = _get_img_iface(line, unk); + static WCHAR sNameImage[] = {'W','i','n','e','I','m','g',0}; Please make it more generic and avoid using lstrcmpW by getting expected name as const char* argument and using strcmp_wa to compare strings. Jacek

1 0

Re: [3/6]user32/tests: Test for SwitchDesktop API
by James McKenzie Dec. 13, 2009

Dec. 13, 2009

shanmukha sainath addepalli wrote: > Hi, > > The test for SwitchDesktop has not submitted earlier. This test > performs switching between two desktops using SwitchDesktop API. > Will this test and the other test compile into winetest_latest.exe and will this test give the results you expect on any version of Windows (95 through 7)? These test should be added first before adding any code to the existing Wine base and validated to have correct values. James McKenzie

1 0