February 2010 - Wine-Devel - list.winehq.org

Re: [1/3] gdiplus: Implement GdipSetImageAttributesGamma.

by Paul Vriens

On 01/31/2010 02:03 AM, Vincent Povirk wrote: > + stat = GdipBitmapGetPixel(bitmap2, 0, 0,&color); > + expect(Ok, stat); > + todo_wine expect(0xff20ffff, color); Hi Vincent, This one introduces a test failure on what appears to be mostly VMware boxes. Not all VMware boxes show this failure btw and I did see the same failure on a real WinMe box. Would a color_match (like we use in several other tests) be more appropriate or maybe a broken()? -- Cheers, Paul.

Feb. 2, 2010

Re: [PATCH] userenv: Implement DestroyEnvironmentBlock().

by Maarten Lankhorst

Hi Lei, 2010/2/2 Lei Zhang <thestig(a)google.com>: > Hi, > > Over the weekend, someone wrote a Chromium base unit test that calls > DestroyEnvironmentBlock(). So I guess it's time to implement it. The > modified Wine test case passed on my WinXP machine. Are you sure that DestroyEnvironmentBlock needs that null pointer check? ~Maarten

Feb. 2, 2010

MS installer frustration: "compressed file wasn't extracted"

by Dan Kegel

Now that plain old Visual C++ 2005 installs pretty well, I've been trying to get Visual C++ 2005 sp1 and Visual C++ 2008 to install. The latter requires .net 3.5. And all of them appear to be blocked by "err:msi:ACTION_InstallFiles compressed file wasn't extracted", http://bugs.winehq.org/show_bug.cgi?id=20110 Grrr. On the bright side, with Jaun's fix today, .net 3.0's runtime installer seems to kinda work.

Feb. 2, 2010

New winetricks 20100201: new verbs d3dx10, directx-beta, ddr=, mspaint, multisampling=, rtlm=, sound=disabled, usp10, vc2005hotfix, vc2005load/vc2005save,

by Dan Kegel

Oh, my, we've been busy. Lots and lots of improvements. Runs better on Windows, too. Cygwin is kinda broken with current wine, so we've rewound back to an old version, but you have to copy and paste a URL from winetricks' output into the mirror selector to use it. Sigh. 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.) Dan Kegel: Add AF's speedup to dotnet30 verb; unpack dotnet35 before running its setup. Remove trailing CR in programfilesdir_win; add vc2005save and vc2005load verbs fix glaring thinko in how we decided whether to run gksudo (i.e. it was horribly broken) Quote more paths Abort if sudo or gksudo fails Work around strange NULs in argument to zenity Internatinalize VirtualCloneDrive code. Support caching cygwin packages (manually for now) Finished vc2005hotfix verb Fixed WINETRICKS_CACHE_WIN value on XP Make vc2005trial verb work on XP Make vcrun2005 work on windows. vc2005trial verb returned before install was really done, tsk. Fix bad font check Work around wine bug 21206, fix cygwin for now by using an old copy. Fix calls to try_regedit to also work on Windows Fix append_path to work on both Windows and Linux (though maybe not solaris?) Austin English: Update shockwave sha1sum Add d3dx10 verb Fix icodec url Add native MSPaint verb. Woohoo stick figures! Fix permission checkign on symlink'ed $HOME's (issue 88) Add option to disable sound (Issue 87) Bump firefox to 3.5.6 Add multisampling enable/disable options Add directdrawrenderer gdi/opengl options Add RenderTargetLockMode override options winetrickstest: move vc2005express to broken, add a note in winetricks Add directx-beta verb (winver=XP, not 2k). Fixes winezeug 71. Andrew Nguyen: Suppress error output from xlsfonts. Hans Leidekker: Add usp10 verb

Feb. 2, 2010

Re: [PATCH 4/5] d3drm: Add some tests

by Christian Costa

Paul Vriens a écrit : > > On 02/01/2010 08:56 AM, Stefan Dösinger wrote: >> On Monday 01 February 2010 08:18:42 Christian Costa wrote: >>> +void Test(void) >>> +{ >>> ... >>> Test(); >> I think the function name Test() won't be too helpful once more tests >> are >> added. You can also make it static, unless you plan to call it from >> another .c >> file. >> >> > > +#define D3DRM_GET_PROC(func) \ > + p ## func = (void*)GetProcAddress(d3drm_handle, #func); \ > + if(!p ## func) { \ > + trace("GetProcAddress(%s) failed\n", #func); \ > > A win_skip() would probably be better here. > > + FreeLibrary(d3drm_handle); \ > + return FALSE; \ > + } > + > +static BOOL InitFunctionPtrs(void) > +{ > + d3drm_handle = LoadLibraryA("d3drm.dll"); > + > + if(!d3drm_handle) > + { > + skip("Could not load d3drm.dll\n"); > > And here as well. > Well. It comes from vector.c. I never use win_skip so far but I can take a look at it. Thanks.

Feb. 1, 2010

Re: [PATCH 4/5] d3drm: Add some tests

by Christian Costa

Stefan Dösinger a écrit : > On Monday 01 February 2010 08:18:42 Christian Costa wrote: > >> +void Test(void) >> +{ >> ... >> Test(); >> > I think the function name Test() won't be too helpful once more tests are > added. You can also make it static, unless you plan to call it from another .c > file. > > Yeap. Right. I will send a patch for that. Thanks.

Feb. 1, 2010

Re: msi: Fix crash with access to component

by James Hawkins

2010/1/30 André Hentschel <nerv(a)dawncrow.de>: > Hans Leidekker schrieb: >> On Saturday 30 January 2010 12:55:21 André Hentschel wrote: >> >>> --- >>> dlls/msi/files.c | 2 ++ >>> 1 files changed, 2 insertions(+), 0 deletions(-) >>> >>> diff --git a/dlls/msi/files.c b/dlls/msi/files.c >>> index 7e74dca..afeb451 100644 >>> --- a/dlls/msi/files.c >>> +++ b/dlls/msi/files.c >>> @@ -363,6 +363,8 @@ static UINT ITERATE_DuplicateFiles(MSIRECORD *row, LPVOID param) >>> { >>> TRACE("Skipping copy due to disabled component %s\n", >>> debugstr_w(component)); >>> + if (!comp) >>> + return ERROR_INSTALL_FAILURE; >>> >>> /* the action taken was the same as the current install state */ >>> comp->Action = comp->Installed; >> >> Where do you see that crash? Are you sure ERROR_INSTALL_FAILURE is appropriate? >> >> -Hans > thought twice(and digging into the code some more), maybe returning a success is a better idea. > And writing a test to determine the correct behavior is an even better idea. -- James Hawkins

Feb. 1, 2010

Re: msi: Don't fail if the service to start is already running.

by Hans Leidekker

On Monday 01 February 2010 16:39:38 Paul Vriens wrote: > 'TermService' is not available on my NT and W2K boxes, so using > 'Spooler' would give us greater coverage. > > Do you agree with the attached? Sure, go ahead. -Hans

Feb. 1, 2010

Re: msi: Don't fail if the service to start is already running.

by Paul Vriens

On 02/01/2010 04:39 PM, Paul Vriens wrote: > On 02/01/2010 04:31 PM, Hans Leidekker wrote: >> On Monday 01 February 2010 16:20:56 Paul Vriens wrote: >> >>> On 02/01/2010 09:55 AM, Hans Leidekker wrote: >>>> + >>>> + scm = OpenSCManager(NULL, NULL, SC_MANAGER_ALL_ACCESS); >>>> + ok(scm != NULL, "Failed to open the SC Manager\n"); >>>> + >>>> + service = OpenService(scm, "TermService", SC_MANAGER_ALL_ACCESS); >>>> + ok(service != NULL, "Failed to open TermService\n"); >>> >>> Hi Hans, >>> >>> Win9x/WinMe don't have the notion of services. 'TermService' is not >>> available on all platforms. We could use 'Spooler' and bail out if we >>> get ERROR_SERVICE_DOES_NOT_EXIST? >>> >>> Let me know and I'll create/send a patch. >> >> Win9x is already detected in install.c so lets just use that. It's not a >> problem if TermService is not available everywhere. >> >> -Hans > > 'TermService' is not available on my NT and W2K boxes, so using > 'Spooler' would give us greater coverage. > > Do you agree with the attached? > > One other thing is that we stop the service regardless of whether it was > already started. Not a big issue as I don't suspect anyone running > winetests on their production boxes, but still. > Forget that last remark. -- Cheers, Paul.

Feb. 1, 2010

Re: msi: Don't fail if the service to start is already running.

by Paul Vriens

On 02/01/2010 04:31 PM, Hans Leidekker wrote: > On Monday 01 February 2010 16:20:56 Paul Vriens wrote: > >> On 02/01/2010 09:55 AM, Hans Leidekker wrote: >>> + >>> + scm = OpenSCManager(NULL, NULL, SC_MANAGER_ALL_ACCESS); >>> + ok(scm != NULL, "Failed to open the SC Manager\n"); >>> + >>> + service = OpenService(scm, "TermService", SC_MANAGER_ALL_ACCESS); >>> + ok(service != NULL, "Failed to open TermService\n"); >> >> Hi Hans, >> >> Win9x/WinMe don't have the notion of services. 'TermService' is not >> available on all platforms. We could use 'Spooler' and bail out if we >> get ERROR_SERVICE_DOES_NOT_EXIST? >> >> Let me know and I'll create/send a patch. > > Win9x is already detected in install.c so lets just use that. It's not a > problem if TermService is not available everywhere. > > -Hans 'TermService' is not available on my NT and W2K boxes, so using 'Spooler' would give us greater coverage. Do you agree with the attached? One other thing is that we stop the service regardless of whether it was already started. Not a big issue as I don't suspect anyone running winetests on their production boxes, but still. -- Cheers, Paul.

Feb. 1, 2010