May 2005 - Wine-devel - list.winehq.org

Re: Write problem with MS Office 2003
by Juan Lang 23 May '05

23 May '05

Stefan wrote: > There's a check for this case: > if (cidl && !apidl) > return E_INVALIDARG; > > MSDN doesn't give me a hint what to do if cidl is set to 0. If I > change the if line to "if (!cidl && !apidl)" (so E_INVALIDARG is > returned to office), saving the file works. Should this line be > "if (!cidl || !apidl)", or is there a special case for cidl==0? There is a special case for cidl == 0, some apps expect to be able to call GetAttributesOf with cidl == 0, apidl == NULL. See the patch that introduced it here: http://www.winehq.org/hypermail/wine-patches/2005/05/0156.html The special case, in particular, is to handle SFGAO_VALIDATE, described on the MSDN page for IShellFolder::GetAttributesOf. I'm guessing this bit is what the app writers were reading when they decided to call GetAttributesOf with cidl == 0: "When used with the file system folder, SFGAO_VALIDATE instructs the folder to discard cached properties retrieved by clients of IShellFolder2::GetDetailsEx that may have accumulated for the specified items." The correct fix is to modify SHELL32_GetItemAttributes to clear SFGAO_READONLY if the file is not read-only. --Juan __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new Resources site http://smallbusiness.yahoo.com/resources/

1 0

Re: Blame ExpandEnvironmentStringsA for the failing NT4 test in shlwapi:shreg
by Alexandre Julliard 23 May '05

23 May '05

Richard Cohen <richard(a)daijobu.co.uk> writes: > This is failing because ExpandEnvironmentStringsA is returning twice > the number of characters it should. > > I would like to mark this test (and some others) as todo(nt40), but > the testing framework isn't flexible enough to let me do that (the > only real options seem to be "wine" and "windows"). > > Would a patch to extend the framework to allow a bit mask, something > like todo(win9x|winxp), be acceptable? A todo is supposed to mark something that will need to be addressed later on. Obviously we can't fix the Windows code, so that's not a todo, it's just a behavior of the API that should either be handled by the test, or simply not tested at all if the results are not reliable. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: install launcher for taskmgr
by Alexandre Julliard 23 May '05

23 May '05

Brian Vincent <vinn(a)theshell.com> writes: > like msiexec, taskmgr didn't have have a launcher script It's really a namespace issues, we don't want to pollute the global /usr/bin with more apps than strictly necessary. Unlike msiexec, taskmgr isn't Windows-specific enough IMO to export it this way. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: Fix Sparse Warnings for server/
by Alexandre Julliard 23 May '05

23 May '05

Robert Shearman <rob(a)codeweavers.com> writes: > Changelog: > Fix sparse warnings. sparse should be taught that 0 is a perfectly fine value (and IMO preferable to NULL) for handles. -- Alexandre Julliard julliard(a)winehq.org

3 3

RESEND: Get AdjustWindowRect working better
by Richard Cohen 23 May '05

23 May '05

Richard Cohen wrote: > >> This fixes a longstanding bug whereby winemine would move 1 pixel down >> the screen for every new game. >> > Changelog: - Fix off-by-one in menu height calculation (& therefore AdjustWindowRect) + test

1 0

RASENH and Crypt Functions and ADVAPI32
by Uwe Bonnes 23 May '05

23 May '05

Michael, comments in rsaenh show that you duplicated code from advapi32. With SystemFunction036 there is another duplicated functionality. Is there any reason why you don't use the advapi functions direct? Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

2 1

Re: winedump: Print file offset instead of obscure 'prefix' while dumping data
by Eric Pouech 22 May '05

22 May '05

Dmitry Timoshkov a écrit : > Hello, > > Changelog: > Dmitry Timoshkov <dmitry(a)codeweavers.com> > Print file offset instead of obscure 'prefix' while dumping data. prefix is used to properly indent the output of the dump, so you do want to keep the prefix (this doesn't prevent you from adding the offsets for dumps) A+ -- Eric Pouech

2 1

Re: urlmon: Prevent possible use of freed memory
by Jacek Caban 22 May '05

22 May '05

Troy Rollo wrote: >This fixes Bugzilla bug 2969 <http://bugs.winehq.org/show_bug.cgi?id=2969> > >ChangeLog: > Prevent a URLMonikerImpl from being freed before its work in > BindToStorage is done. > > >------------------------------------------------------------------------ > >Index: dlls/urlmon/umon.c >=================================================================== >RCS file: /home/wine/wine/dlls/urlmon/umon.c,v >retrieving revision 1.52 >diff -u -r1.52 umon.c >--- dlls/urlmon/umon.c 5 May 2005 09:50:46 -0000 1.52 >+++ dlls/urlmon/umon.c 21 May 2005 23:57:36 -0000 >@@ -456,6 +456,13 @@ > *ppvObject = (void *) This->pstrCache; > IStream_AddRef((IStream *) This->pstrCache); > >+ /* We are about to start calling back into the application. It is >+ * possible that the application will release its reference to us in >+ * these callbacks, so we need to add a reference to This to make sure the >+ * URLMonikerImpl is not freed before we reach the end of this method. >+ */ > > The right solution is separating IBinding and IMoniker interfaces. They shouldn't be implemented in the same object. I'll send the patch. Jacek

2 1

Re: urlmon/umon.c: Fix corruption
by Jacek Caban 22 May '05

22 May '05

Uwe Bonnes wrote: >Changelog: > dlls/urlmon/umon.c: URLMonikerImpl_Release() > Release structure elements before Freeing structure > > My patch I sent today fixes this too. > Changelog: > wine/dlls/urlmon/umon.c:URLMonikerImpl_BindToStorage() > Don't use a HeapAllocated pointer for other purposes ... > + if (path) > + HeapFree(GetProcessHeap(), 0, path); You don't need this check. > Changelog: > dlls/wininet/internet.c: CrackURL, dlls/wininet/tests/http.c > Add some escape/unescape cases and code thor that cases ... > + ok(InternetCrackUrlA(TEST_URL3, 0, ICU_DECODE, &urlComponents),"InternetCrackUrl failed with GLE 0x%lx\n",GetLastError()); > + fprintf(stderr,"BON:%*s\n", (int)urlComponents.dwUrlPathLength, urlComponents.lpszUrlPath); > > } You shouldn't use fprintf() here. Use trace() instead. Jacek

1 0

Re: [OLE #90] Marshal Objects & Monikers into the ROT
by Alexandre Julliard 22 May '05

22 May '05

Robert Shearman <rob(a)codeweavers.com> writes: > Changelog: > - Marshal objects & monikers into the ROT. > - Test for this behaviour. This fails make test here: ../../../tools/runtest -q -P wine -M ole32.dll -T ../../.. -p ole32_test.exe.so moniker.c && touch moniker.ok fixme:ole:MkParseDisplayName (0x77c518a0, L"clsid:20D04FE0-3AEA-1069-A2D8-08002B30309D:", 0x77acfe68, (nil)): stub. err:ole:CoUnmarshalInterface IMarshal::UnmarshalInterface failed, 0x80004005 moniker.c:145: Test failed: Number of monikers should be equal to 2 not 0 err:ole:CoUnmarshalInterface IMarshal::UnmarshalInterface failed, 0x80004005 moniker.c:164: Test failed: Number of monikers should be equal to 2 not 0 fixme:ole:CreateClassMoniker {0002e005-0000-0000-c000-000000000046} err:ole:get_unmarshaler_from_stream Failed to read common OBJREF header, 0x00000001 make[3]: *** [moniker.ok] Error 2 -- Alexandre Julliard julliard(a)winehq.org

2 1