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

Re: shell32/tests: add tests for the parser of SHELLEXECUTEINFO.lpFile [try 2]
by Ilya Basin 17 Feb '10

17 Feb '10

Hi! Please comment on "try 2" of the patch. Will you accept it? IB> Hi! This should expose this bug: IB> http://bugs.winehq.org/show_bug.cgi?id=19385 ( the 'wine start' IB> launcher does not open MS Office documents that have spaces in their IB> path ). Although the title is misleading, many other tickets are IB> marked as duplicates of it. IB> It's not about spaces. It's about sometimes unneeded parsing of IB> SHELLEXECUTEINFO.lpFile, trying to extract arguments from it. Why? IB> Or maybe some other part of wine relies on this behavior? IB> Succeeds on 98 se, 2k & xp. IB> 4 tests fail on wine IB> --- IB> dlls/shell32/tests/shlexec.c | 87 ++++++++++++++++++++++++++++++++++++++++++ IB> 1 files changed, 87 insertions(+), 0 deletions(-) IB> diff --git a/dlls/shell32/tests/shlexec.c b/dlls/shell32/tests/shlexec.c IB> index c673f0a..d575d22 100644 IB> --- a/dlls/shell32/tests/shlexec.c IB> +++ b/dlls/shell32/tests/shlexec.c IB> @@ -797,6 +797,10 @@ static const char* testfiles[]= IB> "%s\\test file.sde", IB> "%s\\test file.exe", IB> "%s\\test2.exe", IB> + "%s\\simple.shlexec", IB> + "%s\\drawback_file.noassoc", IB> + "%s\\drawback_file.noassoc foo.shlexec", IB> + "%s\\drawback_nonexist.noassoc foo.shlexec", IB> NULL IB> }; IB> IB> @@ -852,6 +856,88 @@ static filename_tests_t noquotes_tests[]= IB> {NULL, NULL, 0} IB> }; IB> IB> +static void test_lpFile_parsed(void) IB> +{ IB> + /* basename tmpdir */ IB> + const char* shorttmpdir; IB> + IB> + const char *testfile; IB> + char fileA[MAX_PATH]; IB> + IB> + int rc; IB> + int expected; IB> + IB> + HMODULE hkernel32 = GetModuleHandle("kernel32"); IB> + BOOL isreallyXP2k3plus = NULL != GetProcAddress(hkernel32, "AttachConsole"); IB> + IB> + GetTempPathA(sizeof(fileA), fileA); IB> + shorttmpdir = tmpdir + strlen(fileA); IB> + IB> + /* ensure tmpdir is in %TEMP%: GetTempPath() can succeed even if TEMP is undefined */ IB> + SetEnvironmentVariableA("TEMP", fileA); IB> + IB> + #define TEST_LPFILE_PARSED_OK_() (rc==expected || rc>32 && expected>32) IB> + #define TEST_LPFILE_PARSED_OK() ok(TEST_LPFILE_PARSED_OK_(), "expected %s (%d), got %s (%d), lpFile: %s \n", expected==33 ? "success" : "failure", expected, rc > 32 ? "success" : "failure", rc, fileA) IB> + IB> + /* existing "drawback_file.noassoc" prevents finding "drawback_file.noassoc foo.shlexec" on wine */ IB> + testfile = "%s\\drawback_file.noassoc foo.shlexec"; IB> + expected = 33; sprintf(fileA, testfile, tmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + todo_wine { TEST_LPFILE_PARSED_OK(); } IB> + IB> + /* if quoted, existing "drawback_file.noassoc" not prevents finding "drawback_file.noassoc foo.shlexec" on wine */ IB> + testfile = "\"%s\\drawback_file.noassoc foo.shlexec\""; IB> + expected = 33; sprintf(fileA, testfile, tmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + TEST_LPFILE_PARSED_OK(); IB> + IB> + /* error should be 2, not 31 */ IB> + testfile = "\"%s\\drawback_file.noassoc\" foo.shlexec"; IB> + expected = 2; sprintf(fileA, testfile, tmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + TEST_LPFILE_PARSED_OK(); IB> + IB> + /* ""command"" not works on wine (and real win9x and w2k) */ IB> + if (isreallyXP2k3plus) { IB> + testfile = "\"\"%s\\simple.shlexec\"\""; IB> + expected = 33; sprintf(fileA, testfile, tmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + todo_wine { TEST_LPFILE_PARSED_OK(); } IB> + } else { IB> + win_skip("ShellExecute('\"\"command\"\"', ...) only works on XP/2k3 or newer\n"); IB> + } IB> + IB> + /* nonexisting "drawback_nonexist.noassoc" not prevents finding "drawback_nonexist.noassoc foo.shlexec" on wine */ IB> + testfile = "%s\\drawback_nonexist.noassoc foo.shlexec"; IB> + expected = 33; sprintf(fileA, testfile, tmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + TEST_LPFILE_PARSED_OK(); IB> + IB> + /* is SEE_MASK_DOENVSUBST default flag? Should only be when XP emulates 9x (XP bug or real 95 or ME behavior ?) */ IB> + testfile = "%%TEMP%%\\%s\\simple.shlexec"; IB> + expected = 2; sprintf(fileA, testfile, shorttmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + todo_wine { TEST_LPFILE_PARSED_OK(); } IB> + IB> + /* quoted */ IB> + testfile = "\"%%TEMP%%\\%s\\simple.shlexec\""; IB> + expected = 2; sprintf(fileA, testfile, shorttmpdir); IB> + rc=shell_execute(NULL, fileA, NULL, NULL); IB> + todo_wine { TEST_LPFILE_PARSED_OK(); } IB> + IB> + /* test SEE_MASK_DOENVSUBST works */ IB> + testfile = "%%TEMP%%\\%s\\simple.shlexec"; IB> + expected = 33; sprintf(fileA, testfile, shorttmpdir); IB> + rc=shell_execute_ex(SEE_MASK_DOENVSUBST | SEE_MASK_FLAG_NO_UI, NULL, fileA, NULL, NULL); IB> + TEST_LPFILE_PARSED_OK(); IB> + IB> + /* quoted */ IB> + testfile = "\"%%TEMP%%\\%s\\simple.shlexec\""; IB> + expected = 33; sprintf(fileA, testfile, shorttmpdir); IB> + rc=shell_execute_ex(SEE_MASK_DOENVSUBST | SEE_MASK_FLAG_NO_UI, NULL, fileA, NULL, NULL); IB> + TEST_LPFILE_PARSED_OK(); IB> +} IB> + IB> static void test_filename(void) IB> { IB> char filename[MAX_PATH]; IB> @@ -1938,6 +2024,7 @@ START_TEST(shlexec) IB> IB> init_test(); IB> IB> + test_lpFile_parsed(); IB> test_filename(); IB> test_find_executable(); IB> test_lnks(); --

2 2

Re: [PATCH] d3d8: Display FIXME only once in ValidatePixelShader and ValidateVertexShader
by Henri Verbeet 17 Feb '10

17 Feb '10

On 17 February 2010 11:47, Christian Costa <titan.costa(a)wanadoo.fr> wrote: > I'm not much of a fan of the whole "let's disable the scary debug output" idea, especially since now the function becomes invisible after the first call for people doing actual debugging, unless you run with +relay. WINEDEBUG="-all" is much more effective at shutting up debug output, and tends to have minor performance advantages as well.

2 1

Re: [PATCH] wined3d: Display state in state_undefined
by Henri Verbeet 17 Feb '10

17 Feb '10

On 16 February 2010 11:49, Christian Costa <titan.costa(a)wanadoo.fr> wrote: > - ERR("Undefined state.\n"); > + ERR("Undefined state %d\n", state); Did you test this? "state" is supposed to be always 0. Also, debug_d3dstate() is much more readable than the state number.

3 11

Re: [jscript]add korean translation(resend)
by Paul Vriens 17 Feb '10

17 Feb '10

On 02/16/2010 01:25 AM, Hwang YunSong(황윤성) wrote: > > I append patch file > > <http://www.dreamwiz.com/> http://www.dreamwiz.com/ > > <http://dreamwiz.gmarket.co.kr/redirect2.dwz?http://www.gmarket.co.kr/challe…> > <http://banner.auction.co.kr/bn_redirect.asp?ID=BN00008749&lpurl=http://corn…> > > --- a/dlls/jscript/Makefile.in 2010-02-12 10:23:39.603312863 +0900 +++ b/dlls/jscript/Makefile.in 2010-02-10 09:58:16.684464491 +0900 @@ -13,6 +13,7 @@ jscript_Nl.rc \ jscript_Pt.rc \ jscript_Ru.rc \ + jscript_Ko.rc \ rsrc.rc But you didn't fix the sorting in Makefile.in. -- Cheers, Paul.

1 0

Re: [PATCH] configure: Check for the -Wno-force-align-arg-pointer compiler switch.
by Charles Davis 17 Feb '10

17 Feb '10

Charles Davis wrote: > I added this to clang yesterday. Now clang will warn every time it > encounters the force_align_arg_pointer attribute applied to a function > pointer. This isn't a problem on Linux, but on Mac OS X those warnings > are going to clutter the output, since we use it so liberally. All right, since you're not convinced this is a good idea, let me, well, convince you. The conversation I had on the cfe-commits mailing list: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20100208/027361.… (my original commit) http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20100208/027392.… (someone suggesting the warning) http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20100208/027395.… (someone else seconding it) http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20100208/027396.… (me saying no) http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20100208/027397.… (a third person suggesting the switch) http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20100208/027420.… (me caving) An incomplete list of projects that use force_align_arg_pointer: - liboil - ffmpeg - x264 - Gallium3D - KDE Just do a Google search for 'force_align_arg_pointer' and you'll see a whole bunch of pages about this! Now are you convinced this is a good idea? Chip

2 2

Re: [PATCH] shell32: replace strcpy by memcpy due to slightly undefined struct handling
by Maarten Lankhorst 16 Feb '10

16 Feb '10

Hello Marcus, 2010/2/15, Marcus Meissner <meissner(a)suse.de>: > Hi, > > Writing over the end of a struct via char[1] is only well defined if the > struct > is standalone and not contained in another struct. > > So for gcc 4.5 the strcpy triggers a fortify overflow error and we > better use memcpy. > > Ciao, Marcus > --- > dlls/shell32/pidl.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/dlls/shell32/pidl.c b/dlls/shell32/pidl.c > index 3e61144..a7704b3 100644 > --- a/dlls/shell32/pidl.c > +++ b/dlls/shell32/pidl.c > @@ -1606,7 +1606,7 @@ LPITEMIDLIST _ILCreateEntireNetwork(void) > LPPIDLDATA pData = _ILGetDataPointer(pidlOut); > > pData->u.network.dummy = 0; > - strcpy(pData->u.network.szNames, "Entire Network"); > + memcpy(pData->u.network.szNames, "Entire Network", sizeof("Entire > Network")+1); This is wrong, sizeof() already takes the nul character in account, either use sizeof() or strlen() + 1. Cheers, Maarten.

1 0

Re: mciwave: Stub a boundary case in Save.
by Dmitry Timoshkov 16 Feb '10

16 Feb '10

<Joerg-Cyril.Hoehle(a)t-systems.com> wrote: > This will allow me to enable one more testcase when I'll eventually > submit the long awaited update to mciwave tests. > - if (dwFlags & MCI_WAIT) > - { > - FIXME("MCI_WAIT not implemented\n"); > + if (wmw->hFile == 0) { > + FIXME("Save file with zero-length sample\n"); > + return MCIERR_INTERNAL; > } > - WAVE_mciStop(wDevID, 0, NULL); > + WAVE_mciStop(wDevID, MCI_WAIT, NULL); Please don't change the indentation style of the code. Also, it's not clear to me why you have removed the MCI_WAIT fixme, added MCI_WAIT to mciStop, and treat hFile == 0 as a zero-length sample. -- Dmitry.

2 1

Re: [PATCH 1/4] msctf: Add ITfLangBarMgr Interface [Try 3]
by Alexandre Julliard 16 Feb '10

16 Feb '10

Justin Chevrier <jchevrier(a)gmail.com> writes: > @@ -1357,3 +1362,52 @@ interface ITfThreadFocusSink : IUnknown > > HRESULT OnKillThreadFocus(); > }; > + > +[ > + object, > + uuid(95380031-955c-4710-a2ed-4a2fafe17fe8), > + pointer_default(unique) > +] > +interface ITfLangBarMgr: IUnknown This isn't declared in msctf.idl in the PSDK. -- Alexandre Julliard julliard(a)winehq.org

3 4

Adding wikipedia links to appdb?
by Dan Kegel 16 Feb '10

16 Feb '10

I often want to learn more about an app while I'm in the appdb, and particularly, I want to know whether it has a good reputation. Wikipedia can often tell me what I need to know. It might be handy to have a wikipedia url field in the appdb, and an option in Browse Apps that says "only show apps with wikipedia links". What do folks think about the idea?

7 8

Re: [PATCH 1/3] user32/winex11: Move 32-bit cursor/icon structure to 32-bit header.
by Alexandre Julliard 16 Feb '10

16 Feb '10

Erich Hoover <ehoover(a)mines.edu> writes: > --- a/include/winuser.h > +++ b/include/winuser.h > @@ -98,6 +98,17 @@ typedef HANDLE HDWP; > #define WSF_VISIBLE 1 > #define DF_ALLOWOTHERACCOUNTHOOK 1 > > +typedef struct tagWINECURSORICON { > + POINT ptHotSpot; > + WORD nWidth; > + WORD nHeight; > + WORD nWidthBytes; > + BYTE bPlanes; > + BYTE bBitsPerPixel; > + DWORD frame_size_and; > + DWORD frame_size_xor; > +} WINECURSORICON; You can't do that, private structures have nothing to do in public header files. They also can't be part of the user driver API. -- Alexandre Julliard julliard(a)winehq.org

3 3