November 2008 - Wine-Devel - list.winehq.org

wineconsole: Wait on events in curses backend using select instead of poll
by Kirill K. Smirnov Nov. 29, 2008

Nov. 29, 2008

>Should I add a >check for sys/select.h in the configure script and conditionally included >it, surrounded by HAVE_SYS_SELECT_H? Or would sys/time.h, sys/types.h and >unistd.h be enough to consider it reasonably compatible? Ideally - should. But... 1) There isn't any 'select.h' substring in whole wine tree. 2) There is a file libs/port/usleep.c which uses select() but it does not include any meaningful headers except "config.h" and "wine/port.h". Are these two headers enough wineconsole - I have no idea. Alexandre knows better ;-) Try and see ;-) >After fixing this, I guess I should just submit a new patch against the >current upstream git (not against the old patched version), with the same >commit message, replacing the old one? Yes, and mark a new mail subject with tag '[try 2]' to easily distinguish between different version of the same fix. -- Kirill

2 1

wineconsole: Wait on events in curses backend using select instead of poll
by Kirill K. Smirnov Nov. 29, 2008

Nov. 29, 2008

Hi, Martin, >+ ret = select(pipefd + 1, &readfds, &writefds, &errorfds, NULL); Passing NULL instead of &writefds in this case is better, IMO. It is not used anyway. ...and... I strongly believe that errorfds parameter os select() is useful for sockets only, not for stdio (I am not sure whether it is true for Darwin). Please, check this. -- Kirill

2 1

Implementing wtmpnam in msvcrt
by Tim Holy Nov. 29, 2008

Nov. 29, 2008

Hello, I'm trying to implement the "stub" function wtmpnam in msvcrt. Below is a patch at an attempt; there are a number of questions I have about the right way to implement it, and since I don't know the Wine codebase I thought perhaps readers of this list might (with minimal effort) be able to point me in the right direction. Below is the text of the patch (it's short). My questions are embedded after //-style comments (4 in total). Many thanks in advance for your help! --Tim diff -U 3 -r winecopy/wine-1.1.8/dlls/msvcrt/file.c wine-1.1.8/dlls/msvcrt/file.c --- winecopy/wine-1.1.8/dlls/msvcrt/file.c 2008-11-07 10:09:33.000000000 -0600 +++ wine-1.1.8/dlls/msvcrt/file.c 2008-11-29 07:32:43.000000000 -0600 @@ -88,6 +88,7 @@ /* INTERNAL: Static buffer for temp file name */ static char MSVCRT_tmpname[MAX_PATH]; +static MSVCRT_wchar_t MSVCRT_wtmpname[MAX_PATH]; static const unsigned int EXE = 'e' << 16 | 'x' << 8 | 'e'; static const unsigned int BAT = 'b' << 16 | 'a' << 8 | 't'; @@ -2990,6 +2991,32 @@ } /********************************************************************* + * wtmpnam (MSVCRT.@) + */ +MSVCRT_wchar_t * CDECL MSVCRT_wtmpnam(MSVCRT_wchar_t *s) +{ + static int unique; + char tmpstr[16]; + MSVCRT_wchar_t wtmpstr[16]; + MSVCRT_wchar_t *p; + int count; + if (s == 0) + s = MSVCRT_wtmpname; + msvcrt_int_to_base32(GetCurrentProcessId(), tmpstr); + p = s + MSVCRT_swprintf(s, "\\s%s.", tmpstr); // "implicit declaration" error, function is defined in wcs (but no wcs.h); but how to convert format string to MSVCRT_wchar_t? + for (count = 0; count < MSVCRT_TMP_MAX; count++) + { + msvcrt_int_to_base32(unique++, tmpstr); + mbstowcs(wtmpstr,tmpstr,strlen(tmpstr)); // need a MSVCRT_wchar version + wcscpy(p, wtmpstr); // need a MSVCRT_wchar version + if (GetFileAttributesW(s) == INVALID_FILE_ATTRIBUTES && + GetLastError() == ERROR_FILE_NOT_FOUND) + break; + } + return s; +} + +/********************************************************************* * tmpfile (MSVCRT.@) */ MSVCRT_FILE* CDECL MSVCRT_tmpfile(void) diff -U 3 -r winecopy/wine-1.1.8/dlls/msvcrt/msvcrt.spec wine-1.1.8/dlls/msvcrt/msvcrt.spec --- winecopy/wine-1.1.8/dlls/msvcrt/msvcrt.spec 2008-11-07 10:09:33.000000000 -0600 +++ wine-1.1.8/dlls/msvcrt/msvcrt.spec 2008-11-29 07:12:10.000000000 -0600 @@ -579,7 +579,7 @@ @ cdecl _wstrtime(ptr) @ cdecl _wsystem(wstr) @ cdecl _wtempnam(wstr wstr) -@ stub _wtmpnam #(ptr) +@ cdecl wtmpnam(ptr) MSVCRT_wtmpnam // needs an underscore? @ cdecl _wtoi(wstr) ntdll._wtoi @ cdecl _wtoi64(wstr) ntdll._wtoi64 @ cdecl _wtol(wstr) ntdll._wtol

3 6

new winetricks-20081129: new verbs mshflxgd, controlpad/fm20, native_oleaut32
by Dan Kegel Nov. 29, 2008

Nov. 29, 2008

Also clean up temp directory between verbs, as some verbs were writing readonly files with unique names like 'setup.exe' and stumbling over each other. As always, online at http://kegel.com/wine/winetricks Please submit bug reports to http://code.google.com/p/winezeug/issues/list

1 0

Re: winex11.drv: change bpp fixme to a warn
by James Hawkins Nov. 28, 2008

Nov. 28, 2008

On Fri, Nov 28, 2008 at 4:25 PM, Austin English <austinenglish(a)gmail.com> wrote: > This isn't something we can really fix on our end, so a FIXME isn't appropriate. > diff --git a/dlls/advapi32/security.c b/dlls/advapi32/security.c index 8229377..16aa8c3 100644 --- a/dlls/advapi32/security.c +++ b/dlls/advapi32/security.c @@ -1155,6 +1155,7 @@ DWORD WINAPI BuildSecurityDescriptorW( BOOL WINAPI InitializeSecurityDescriptor( PSECURITY_DESCRIPTOR pDescr, DWORD revision ) { + pDescr = 0; return set_ntstatus( RtlCreateSecurityDescriptor(pDescr, revision )); } Extraneous change. -- James Hawkins

2 1

Re: shell32/autocomplete: fix handling of WM_KILLFOCUS messages
by Austin English Nov. 28, 2008

Nov. 28, 2008

On Fri, Nov 28, 2008 at 10:53 AM, Jan de Mooij <jandemooij(a)gmail.com> wrote: > In eMule after moving focus from the searchbox to another control, it > still shows the caret and search is unusable without switching tabs. > > Autocomplete overrides the textbox message handler. But after handling > WM_KILLFOCUS messages we should send it to the old handler so it can > move the focus from the textbox, hide the caret, etc. > > Fixes bug 4090: http://bugs.winehq.org/show_bug.cgi?id=4090 > > Regards > > > > Can you add a testcase for this? -- -Austin

1 0

Re: dinput: Remove WINAPI on static functions where not needed.
by Vitaliy Margolen Nov. 28, 2008

Nov. 28, 2008

Francois Gouget wrote: > --- > > If you know other places where the functions touched by this patch > are used, please let me know. For more details, see: > > http://www.winehq.org/pipermail/wine-patches/2008-November/064995.html > > > dlls/dinput/joystick_linux.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/dlls/dinput/joystick_linux.c b/dlls/dinput/joystick_linux.c > index 4146f1d..47c734d 100644 > --- a/dlls/dinput/joystick_linux.c > +++ b/dlls/dinput/joystick_linux.c > @@ -966,7 +966,7 @@ static HRESULT WINAPI JoystickAImpl_GetProperty( > /****************************************************************************** > * GetObjectInfo : get object info > */ > -static HRESULT WINAPI JoystickWImpl_GetObjectInfo(LPDIRECTINPUTDEVICE8W iface, > +static HRESULT JoystickWImpl_GetObjectInfo(LPDIRECTINPUTDEVICE8W iface, > LPDIDEVICEOBJECTINSTANCEW pdidoi, DWORD dwObj, DWORD dwHow) > { > static const WCHAR axisW[] = {'A','x','i','s',' ','%','d',0}; Please don't apply this patch. This function should be used in vtable but isn't by mistake. Sending a patch to fix this. Vitaliy.

1 0

Re: [1/2] [try 4] mlang: Add implementation offnIMLangFontLink_GetCharCodePages
by Dmitry Timoshkov Nov. 28, 2008

Nov. 28, 2008

"Vitaly Perov" <vitperov(a)etersoft.ru> wrote: > Changes since last send: > - Used local mlang_data structure instead of calling internal function > wine_cp_enum_table. > - Used macro instead of "iface->lpVtbl->...." Looks good to me. -- Dmitry.

2 1

New winetricks 20081127: new verbs mono20, ie6, msmask, ole2
by Dan Kegel Nov. 27, 2008

Nov. 27, 2008

Also updates to latest directx9, and fixes a Mac problem. As always, online at http://kegel.com/wine/winetricks Please submit bug reports to http://code.google.com/p/winezeug/issues/list Thanks to everybody who sent in suggestions and patches! ie6 - Steven Edwards msmask - Austin mac bugfix - Ryan Schmidt

1 0

Re: programs: Fix broken compilation on win64
by Rob Shearman Nov. 27, 2008

Nov. 27, 2008

2008/11/27 Maarten Lankhorst <maarten(a)codeweavers.com>: > --- > Set the Spare[0] at runtime after a critical section is entered. :-) > > I'm not testing Spare[0] at all calls to EnterCriticalSection, but if this > is a problem I will send a revised patch IMHO, this is too fragile. My solution would be two macros: one for declaring a variable storing the name for a critical section, and a second for putting that name into the critical section structure. The latter would be something like the following for 64-bit: #define WINE_DEBUG_USE_CRIT_SECT_NAME(name) { (DWORD)(DWORD_PTR)name, (DWORD)((DWORD_PTR)name >> 32) } It might even be better to let the macro fill in / declare the entire CRITICAL_SECTION_DEBUG stucture. -- Rob Shearman

2 1

Re: d3dxof: DPRINTF -> TRACE.
by Christian Costa Nov. 27, 2008

Nov. 27, 2008

DPRINTF use here is intentional. Christian Andrew Riedi a écrit : > Changelog: > d3dxof: DPRINTF -> TRACE. > > > ------------------------------------------------------------------------ > >

1 0

Re: [3/5] WineD3D: Fix D3DFMT_R32F, R16F, G16R16F, ... in the pixel shader
by Roderick Colenbrander Nov. 27, 2008

Nov. 27, 2008

Hi, I don't know much about the current shader code but is this the right way to go thinking about GL_EXT_texture_swizzle + GL_ARB_texture_rg? For instance Nvidia drivers will soon get the swizzle extension and for that you don't have to fixups anymore. Is moving this way the right approach, so that it won't become a mess later on? Roderick -- Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger

3 11

Re: [1/2] [try 3] mlang: Add implementation offnIMLangFontLink_GetCharCodePages
by Dmitry Timoshkov Nov. 27, 2008

Nov. 27, 2008

"Vitaly Perov" <vitperov(a)etersoft.ru> wrote: > Changes since last send: > - Used WINAPI function instead of accessing Wine internal code page > structures. You still access internal Wine code page structures, I'm not sure why you claim otherwise. > "Dmitry Timoshkov" <dmitry(a)codeweavers.com> wrote: >>It *is* possible to use EnumSystemCodePages > > Yes, it is possible. But : > 1) this function call wine_cp_enum_table, then converts the result (codepage) > to the string. So we need convert the result back to the number. > 2) It pass code pages to the callback-function. The callback function uses > only one argument - string with codepage number. So it's very difficult to > pass this value back to the GetCharCodePages function. > > So, I think the best decision is to use internal function wine_cp_enum_table What's happend to the idea to use mlang_data? And again: forget about Wine internal code page structures, that's the implementation detail that's internal to kernel32/ntdll, that structures should not be used outside of kernel32/ntdll. > + iface->lpVtbl->CodePageToCodePages(iface, > + cpTable->info.codepage, &codePages); Have a look how other parts in Wine call vtable members, the tests have enough code samples. -- Dmitry.

1 0

Re: user32: Improve the LoadImage() test.
by Dmitry Timoshkov Nov. 27, 2008

Nov. 27, 2008

"Andrew Riedi" <andrewriedi(a)gmail.com> wrote: > - icon_entry->xHotspot = 1; > - icon_entry->yHotspot = 1; > + icon_entry->xHotspot = 1; /* Color planes for .ico. */ > + icon_entry->yHotspot = ICON_BPP; /* BPP for .ico. */ ... > ok(icon_info.xHotspot == 1, "xHotspot is %u.\n", icon_info.xHotspot); > - ok(icon_info.yHotspot == 1, "yHotspot is %u.\n", icon_info.yHotspot); > + ok(icon_info.yHotspot == 32, "yHotspot is %u.\n", icon_info.yHotspot); It should be ICON_BPP instead of 32 for consistency. > + /* Test loading an icon as an icon. */ > + SetLastError(0xdeadbeef); > + handle = LoadImageA(NULL, "icon.ico", IMAGE_ICON, 0, 0, LR_LOADFROMFILE); > + ok(handle != NULL, "LoadImage() failed.\n"); > + error = GetLastError(); > + ok(error == 0 || > + broken(error == 0xdeadbeef) || /* Win9x */ > + broken(error == ERROR_BAD_PATHNAME), /* Win98, WinMe */ > + "Last error: %u\n", error); There is no point in testing last error value if the API didn't fail. > + /* Test the icon information. */ > + SetLastError(0xdeadbeef); > + ret = GetIconInfo(handle, &icon_info); > + ok(ret, "GetIconInfo() failed.\n"); > + error = GetLastError(); > + ok(error == 0xdeadbeef, "Last error: %u\n", error); Same here. > + if (ret) > + { > + ok(icon_info.fIcon == TRUE, "fIcon == FALSE.\n"); > + ok(icon_info.xHotspot == 16, "xHotspot is %u.\n", icon_info.xHotspot); > + ok(icon_info.yHotspot == 16, "yHotspot is %u.\n", icon_info.yHotspot); > + ok(icon_info.hbmColor != NULL, "No hbmColor!\n"); > + ok(icon_info.hbmMask != NULL, "No hbmMask!\n"); > + } 'if (ret)' is useless if you don't expect a failure. > + > + /* Clean up. */ > + SetLastError(0xdeadbeef); > + ret = DestroyIcon(handle); > + ok(ret, "DestroyIcon() failed.\n"); > + error = GetLastError(); > + ok(error == 0xdeadbeef, "Last error: %u\n", error); Same here. -- Dmitry.

3 2

Re: ole32: Add Co*InitializeSpy functions
by Rob Shearman Nov. 27, 2008

Nov. 27, 2008

2008/11/26 Maarten Lankhorst <maarten(a)codeweavers.com>: > + > +HRESULT WINAPI CoRevokeInitializeSpy(ULARGE_INTEGER cookie) > +{ > + struct oletls *info = COM_CurrentInfo(); > + TRACE("(%s)\n", wine_dbgstr_longlong(cookie.QuadPart)); > + > + if (!info || !info->spy || cookie.QuadPart != (DWORD_PTR)info->spy) > + return E_INVALIDARG; > + > + IUnknown_Release(info->spy); > + return S_OK; > +} No function documentation and you forget to clear info->spy before returning. -- Rob Shearman

2 1

Re: [2/5] WineD3D: Use a separate enum for pixel format fixups
by Henri Verbeet Nov. 27, 2008

Nov. 27, 2008

2008/11/27 Stefan Dösinger <stefan(a)codeweavers.com>: > +enum shader_color_conv > +{ > + fmtconv_none = 1, > + set_blue_1 = 2, > + signed_2_no_ba = 3, > + signed_2 = 4, > + signed_4 = 5, > + signed_l6v5u5 = 6, > + swap_rg_blue_1 = 7, > + a_to_g_ba_1 = 8, > + yuv_yuy2 = 9, > + yuv_uyvy = 10, > + yuv_yv12 = 11, > + force_dword = 0x7ffffff > +}; > Could you please use uppercase enum elements like the rest of the world? :-/

1 0

Re: ntdll: DPRINTF -> TRACE. (heap.c)
by Dmitry Timoshkov Nov. 27, 2008

Nov. 27, 2008

"Andrew Riedi" <andrewriedi(a)gmail.com> wrote: > Changelog: > ntdll: DPRINTF -> TRACE. (heap.c) DPRINTF is used for a reason here. -- Dmitry.

2 2

Re: imm32: Add some really basic tests for ImmSetCompositionString().
by Dmitry Timoshkov Nov. 27, 2008

Nov. 27, 2008

"Andrew Riedi" <andrewriedi(a)gmail.com> wrote: > +static int test_ImmSetCompositionString(void) It's a usual practice to have the test cases return 'void', especially if you don't use the return value. > +{ > + HIMC imc; > + BOOL ret; > + DWORD error; > + > + imc = ImmGetContext(hwnd); First thing to test is the return value of ImmGetContext() and report last error in the case of a failure. -- Dmitry.

2 1

Re: [1/5] D3D9: Document a Microsoft header bug
by Adam Petaccia Nov. 26, 2008

Nov. 26, 2008

On Thu, 2008-11-27 at 01:30 +0100, Stefan Dösinger wrote: > > + > + /* Note: Microsoft's d3d9.h does not declare > IDirect3D9Ex::RegisterSoftwareDevice . This would mean that > + * the offsets of the other methods in the Vtable change too. > This, however, seems wrong. In Microsoft's > + * d3d9.dll, the offsets for the other functions are still > compatible with IDirect3D9. > + * (This is probably because even in MS's header IDirect3D9Ex > inherits from IDirect3D9, which makes the > + * C++ inferface compatible, and nobody uses the C interface in > Windows world > + */ Maybe this should be documented via Wine's API doc "BUGS: " declaration as well: http://www.winehq.org/site/docs/winedev-guide/api-docs

1 0

Re: mshtml: Do a case-insensitive compare of type (resend 1)
by Alistair Leslie-Hughes Nov. 26, 2008

Nov. 26, 2008

Any thing wrong with this patch? Best Regards Alistair Leslie-Hughes "Alistair Leslie-Hughes" <leslie_alistair(a)hotmail.com> wrote in message news:492681BB.2090608(a)hotmail.com... > Hi, > > Changelog: > mshtml: Do a case-insensitive compare of type > -------------------------------------------------------------------------------- > >From ab3346563682349645f07147c5adbcfc3ce90ebf Mon Sep 17 00:00:00 2001 > From: Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> > Date: Thu, 13 Nov 2008 21:38:13 +1100 > Subject: [PATCH] Do a case-insensitive compare of type > To: wine-patches <wine-patches(a)winehq.org> > > --- > dlls/mshtml/script.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/dlls/mshtml/script.c b/dlls/mshtml/script.c > index 4d10853..90bfef9 100644 > --- a/dlls/mshtml/script.c > +++ b/dlls/mshtml/script.c > @@ -640,7 +640,7 @@ static BOOL get_guid_from_type(LPCWSTR type, GUID > *guid) > {'t','e','x','t','/','j','a','v','a','s','c','r','i','p','t',0}; > > /* FIXME: Handle more types */ > - if(!strcmpW(type, text_javascriptW)) { > + if(!strcmpiW(type, text_javascriptW)) { > *guid = CLSID_JScript; > }else { > FIXME("Unknown type %s\n", debugstr_w(type)); > -- > 1.5.4.3 > > -------------------------------------------------------------------------------- > >

1 0

re: RFC: Proposed new web site design
by Dan Kegel Nov. 26, 2008

Nov. 26, 2008

Looks great so far. Comments: 1) nested scrollbars are evil. Better to put the news on its own tab, lose the little sidebar the news was in, and have a very small area in the main box for just the one most important headline. 2) The main choice text needs to be closer to user tasks. So instead of Information Overview of Wine features. we should have About What is Wine, and why should I use it? and we're missing the user task Will my app work with Wine? which should link to the appdb.

8 8

Re: atl: Remove WINAPI on static functions where not needed.
by Ken Thomases Nov. 26, 2008

Nov. 26, 2008

Thanks for taking care of this, Francois. On Nov 23, 2008, at 5:15 PM, Francois Gouget wrote: > Astute readers will notice that this leaves one loophole: nasty macro > use like in dlls/winmm/message16.c: > > #define A(_x,_y) MMDRV_InstallMap(_x, \ > MMDRV_##_y##_Map16To32W, MMDRV_##_y##_UnMap16To32W, \ > MMDRV_##_y##_Map32WTo16, MMDRV_##_y##_UnMap32WTo16, \ > MMDRV_##_y##_Callback) > A(MMDRV_AUX, Aux); > > A grep would not find that MMDRV_Aux_Callback is referenced by the > above > code. Could you perhaps grep the output from the preprocessor, instead? -Ken

2 1

Re: mshtml: suppress module unloading while gecko is loaded (take 2)
by Paul Vriens Nov. 26, 2008

Nov. 26, 2008

Dan Kegel wrote: > Thanks to Mike Hearn for pointing out that mshtml reference > counting was wrong, and Jacek for pointing out how to fix > it for now. > > Once gecko is loaded, don't let mshtml unload. > This fixes a crash in starting Sketchup (bug 16164), > and probably fixes a number of other apps that > went pear-shaped after unloading mshtml... > > > ------------------------------------------------------------------------ > > C++ style comment? Was this a Patchwatcher test ;) -- Cheers, Paul.

1 0

Icons as cursors patch
by Krzysztof Drewniak Nov. 25, 2008

Nov. 25, 2008

This is my first patch. It allows icons to be loaded as cursors because it defaults hotspots to x = 0 and y = 0 if the hotspot isn't found. Pasting patch into body of message. This is against the latest git . -------------------------------------------------------------------------------------------------------------------------- --- cursoricon.c.orig 2008-11-23 15:07:32.000000000 -0600 +++ cursoricon.c 2008-11-23 15:13:22.000000000 -0600 @@ -932,7 +932,10 @@ static HICON CURSORICON_LoadFromFile( LP entry = CURSORICON_FindBestIconFile( dir, width, height, colors ); if ( !entry ) - goto end; + /* this is my change. If there is no hotspot, default to 0, 0. */ + hotspot.x = 0; + hotspot.y = 0; + goto iconskip; /* check that we don't run off the end of the file */ if ( entry->dwDIBOffset > filesize ) @@ -942,6 +945,7 @@ static HICON CURSORICON_LoadFromFile( LP hotspot.x = entry->xHotspot; hotspot.y = entry->yHotspot; +iconskip: hIcon = CURSORICON_CreateIconFromBMI( (BITMAPINFO *)&bits[entry->dwDIBOffset], hotspot, !fCursor, 0x00030000, width, height, loadflags ); -------------------------------------------------------------------------------------------------------------------------- Please look over the code. Krzys D.

4 7

Re: mshtml: suppress module unloading until we figure out why some apps crash with it enabled
by Jacek Caban Nov. 25, 2008

Nov. 25, 2008

Dan Kegel wrote: > This prevents http://bugs.winehq.org/show_bug.cgi?id=16164 > > ( http://www.microsoft.com/msj/0197/activex/activex0197.aspx looks > like a nice, scary article on the subject. ) > - return module_ref ? S_FALSE : S_OK; + /* Some apps crash if we let mshtml be unloaded. + * Until we figure out why, don't let mshtml be unloaded. + */ + return S_FALSE; /* module_ref ? S_FALSE : S_OK; */ I think we should prevent unloading mshtml.dll if Gecko was loaded. To do it I'd suggest to increase module count in load_gecko function (it will never be decreased so DLL won't be unloaded). Jacek

1 0