August 2013 - Wine-Devel - list.winehq.org

Re: [PATCH 01/11] widl: New option --rt for enabling rt's specific language extensions.
by André Hentschel Aug. 9, 2013

Aug. 9, 2013

Am 08.08.2013 17:51, schrieb Kai Tietz: > Hi, > > this patch adds new --rt option to widl. By it you can control, if > RT's IDL-language-extension(?s) getting active. > This initial patch just adds the option-scanning and the global > flag-variable to widl. > > Ok for apply? I'd say no, you are adding dead code, so you might want to merge it with 2/11 to avoid that.

3 2

Re: [2/2] wineserver: perform access check when creating event objects (try 2)
by Andrew Cook Aug. 9, 2013

Aug. 9, 2013

Actually ignore this patch, the problem was introduced by 92db6d2c back in 2007, server: Don't do access checks on the security descriptors of newly created objects which fixes sync.c:363 "CreateEventW with blank sd failed" i'll need to take another look at this, calling CreateEventEx with generic flags fails to correctly set permissions, the patch as it stands though will likely cause regressions. On 02/08/13 21:58, Andrew Cook wrote: > --- > dlls/ntdll/tests/om.c | 4 ++-- > server/event.c | 5 +---- > 2 files changed, 3 insertions(+), 6 deletions(-) > >

1 0

Re: configure.ac: allow disabling winemac.drv
by Ken Thomases Aug. 8, 2013

Aug. 8, 2013

On Aug 8, 2013, at 5:56 PM, Austin English wrote: > +AC_ARG_WITH(winemac, AS_HELP_STRING([--without-winemac],[do not build native Mac (Cocoa) driver]), > + [if test "x$withval" = "xno"; then ac_cv_header_ApplicationServices_ApplicationServices_h=no; fi]) A switch to disable the Mac driver is fine with me, but you can't pretend that the ApplicationServices framework header isn't present. There are other places within Wine which check that. http://source.winehq.org/source/dlls/windowscodecs/icnsformat.c#L24 http://source.winehq.org/source/dlls/winspool.drv/info.c#L48 -Ken

1 0

Re: [PATCH 04/11] widl: Handle default value for public vs hidden attribute.
by André Hentschel Aug. 8, 2013

Aug. 8, 2013

Am 08.08.2013 17:57, schrieb Kai Tietz: > - if (! is_attr(attrs, ATTR_PUBLIC)) > + if (! is_attr(attrs, ATTR_PUBLIC) && ! is_attr (attrs, ATTR_HIDDEN)) style issue: space after !

2 1

Re: [PATCH 09/11] widl: Handle LLP64 target same as LP64 one.
by André Hentschel Aug. 8, 2013

Aug. 8, 2013

Am 08.08.2013 18:07, schrieb Kai Tietz: > Hi, > > this thing here puzzles me most. I reject this patch, but would like > to get here some discussion to find the actual intended meaning of > this xstrtoul function. Then you should remove it from the patchset and send it as RFC to wine-devel

1 0

Re: Synchronization-patches of widl
by André Hentschel Aug. 8, 2013

Aug. 8, 2013

Am 08.08.2013 10:56, schrieb Kai Tietz: > All these changes were necessary to support the VLC-people to begin > with their WinRT port. By this we began on mingw-w64 to improve - > thanks to Jacek - our IDL-generated header-sources. So you are on a good way with WinRT, nice. What about Windows RT, was there already some ARM work? > A general note I would like to make about the use of sizeof(void*) in > widl to make out what type-library-variant to be choosen. That seems > to be a bit wacky variant, as it makes widl non-useable as cross-tool > on host with different pointer-size. isn't that controled by command line arguments?

1 0

winetestbot question
by Stefan Leichter Aug. 8, 2013

Aug. 8, 2013

Hello, i like to know why the WXPX64 VM of winetestbot does not have the 32 bit vc runtime 2008. The other 64 bit VMs have the runtime installed. Because of the missing runtime each patch for the unit tests of the dlls msvcr90 and msvcp90 is marked as "Failed" at http://source.winehq.org/patches -- Stefan

2 1

re: msvcrt: strncpy doesn't compliant C standard (try 3)
by Dan Kegel Aug. 8, 2013

Aug. 8, 2013

Minor problem: +static void test_strncpy(void) +{ + size_t len = 10; + char *ret; + char dst[len + 1]; Hmm. That last line is a VLA, and might not compile in all C compilers because it's not allowed in C89. http://stackoverflow.com/questions/448844/variable-sized-arrays-in-c Wine seems to want to build in c89 compilers: http://www.winehq.org/pipermail/wine-cvs/2011-April/076721.html Try using a #define with a more unique name for the length instead, or better yet, replace len=10 with len=sizeof(dst)-1 or something. Larger problem: ret = strncpy(dst, not_null_less, len); This is ill-defined. You're copying 10 bytes from a 5 byte buffer. Don't do that. Also, passing a not-null-terminated string in as a special test case doesn't seem interesting... just pass in the longer null-terminated string, and specify a shorter length. Fewer test strings needed, easier to read code. Please run 'make check', and test using http://testbot.winehq.org, before submitting patches.

2 1

Re: [PATCH 1/1] msvcrt: Set ?: environment variable in _(w)chdir functions.
by Marvin Aug. 8, 2013

Aug. 8, 2013

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=26639 Your paranoid android. === WINEBUILD (build) === Make failed

1 0

Re: msvcrt: strncpy doesn't compliant C standard (try 3)
by Marvin Aug. 7, 2013

Aug. 7, 2013

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=26638 Your paranoid android. === W2KPROSP4 (32 bit string) === string.c:2585: Test failed: dst != 012345 === WXPPROSP2 (32 bit string) === string.c:2585: Test failed: dst != 012345 === WXPX64 (32 bit string) === string.c:2585: Test failed: dst != 012345 === W2K3R2SESP2 (32 bit string) === string.c:2585: Test failed: dst != 012345 === WVISTAX64 (32 bit string) === string.c:2585: Test failed: dst != 012345 === W2K8SE (32 bit string) === string.c:2585: Test failed: dst != 012345 === W7PROX64 (32 bit string) === string.c:2585: Test failed: dst != 012345 === TEST64_W7SP1 (32 bit string) === string.c:2585: Test failed: dst != 012345

1 0

Re: msvcp90/tests: Add dynamically loaded functions for ARM
by Marvin Aug. 7, 2013

Aug. 7, 2013

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=26637 Your paranoid android. === WXPX64 (32 bit misc) === No test summary line found === WXPX64 (32 bit ios) === No test summary line found

1 0

Re: Request to add wow64 keyword to bugzilla
by Ken Sharp Aug. 7, 2013

Aug. 7, 2013

+1 from me. On 07/08/13 15:03, Rosanne DiMesio wrote: > On Wed, 07 Aug 2013 14:23:53 +0100 > Ken Sharp <kennybobs(a)o2.co.uk> wrote: > > >> Would I be right in assuming you would like to see bugs in 32-bit >> applications that are only present in a wow64 WINEPREFIX? Are there many? > > Yes. As to how many there are, I've been able to find 7 by doing searches of summaries/comments for things like "wow64" and "64" and then reading the bugs to see if they fit, but since people have such wildly varying ways of phrasing things, I don't know if that's all of them. I have seen it often enough on the forum that I now routinely tell 64 bit users to try a 32 bit wineprefix and have added instructions on how to create one to the FAQ. That's why I'd like an easier way to track these bugs. I see. A lot of people say this over and over again. +1 from me for a wow64 keyword. They may even be fairly easy to solve (picking the right paths / registry entries). > > As to keyword usage, some of those bugs were tagged with the win64 keyword, some weren't. I know there is also a win32 keyword, and perhaps that is meant to be used for the kinds of bugs I am looking for, but if so, that is not at all clear from the keyword definition, and it is not how people are using the keywords. > There's no win32 keyword?

3 2

Re: msvcp90/tests: Skip tests on missing functions
by Piotr Caban Aug. 7, 2013

Aug. 7, 2013

On 08/06/13 23:00, André Hentschel wrote: > @@ -749,6 +749,13 @@ static BOOL init(void) > > SET(p_basic_istream_char_read_uint64, > "??5?$basic_istream(a)DU?$char_traits(a)D@std@@@std@@QAEAAV01(a)AA_K@Z"); > + > + if (!p_basic_istream_char_read_uint64) > + { > + skip("_basic_istream_char_read_uint64 not found, skipping tests\n"); > + return FALSE; > + } Why are you skipping the tests instead of properly initializing functions pointers on ARM?

2 1

Request to add wow64 keyword to bugzilla
by Rosanne DiMesio Aug. 7, 2013

Aug. 7, 2013

Can wow64 be added as a keyword to bugzilla? I know we already have win64 as a keyword, but that's being used for both 64 bit apps and 32 bit apps in a 64 bit wineprefix. I'm interested in being able to track the latter, as it's hitting increasing numbers of users. -- Rosanne DiMesio <dimesio(a)earthlink.net>

2 1

Re: [PATCH] typo fixes (https://github.com/vlajos/misspell_fixer)
by Frédéric Delanoy Aug. 7, 2013

Aug. 7, 2013

On Mon, Aug 5, 2013 at 9:41 PM, Veres Lajos <vlajos(a)gmail.com> wrote: > > Signed-off-by: Veres Lajos <vlajos(a)gmail.com> > --- > dlls/comctl32/tests/trackbar.c | 2 +- > documentation/ChangeLog.ALPHA | 80 ++++++++++++++++++++-------------------- > documentation/ChangeLog.BETA | 44 +++++++++++----------- > documentation/ChangeLog.OLD | 24 ++++++------ > tools/winebuild/utils.c | 2 +- > 5 files changed, 76 insertions(+), 76 deletions(-) The "documentation/ChangeLog.*" files have been removed so the patch won't apply. Frédéric Delanoy

1 0

Re: msvcrt: strncpy doesn't compliant C standard (try 2)
by Frédéric Delanoy Aug. 7, 2013

Aug. 7, 2013

On Tue, Aug 6, 2013 at 10:44 PM, Álvaro Nieto <alvaro.nieto(a)gmail.com> wrote: > Remove strlen and implement some tests You should probably add tests for when the source string is not null-terminated. Frédéric Delanoy

1 0

Re: msvcp90/tests: Skip tests on missing functions
by Marvin Aug. 6, 2013

Aug. 6, 2013

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=26628 Your paranoid android. === WXPX64 (32 bit misc) === No test summary line found === WXPX64 (32 bit ios) === No test summary line found

1 0

Re: msvcr90/tests: Skip when _type_info_name_internal_method is not found
by Marvin Aug. 6, 2013

Aug. 6, 2013

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=26627 Your paranoid android. === WXPX64 (32 bit msvcr90) === No test summary line found

1 0

Re: ws2_32/tests: Fix a comment
by Juan Lang Aug. 6, 2013

Aug. 6, 2013

Hi Andre, - /* this is a io heavy test, do it at the end so the kernel doesn't start dropping packets */ + /* this is a heavy io test, do it at the end so the kernel doesn't start dropping packets */ To my eyes, this isn't an improvement. A slight improvement might be "this is a io-heavy test", or "this is an io-heavy test", but I'm not sure it's enough of an improvement to bother. --Juan

1 0

Re: msvcrt: strncpy doesn't compliant C standard
by Nikolay Sivov Aug. 6, 2013

Aug. 6, 2013

On 8/6/2013 11:09, Álvaro Nieto wrote: > This patch solves [Bug 34211]. The implementation of strncpy function > doesn't compliant with C standard [1]. Also Microsoft Visual Studio > C/C++ compiler is ok with the standard [2]. > > Extract from msdn; > > "The strncpy function copies the initial count characters of strSource > to strDest and returns strDest. If count is less than or equal to the > length of strSource, a null character is not appended automatically to > the copied string. If count is greater than the length of strSource, > the destination string is padded with null characters up to length > count. The behavior of strncpy is undefined if the source and > destination strings overlap." > > [1] http://pubs.opengroup.org/onlinepubs/009695399/functions/strncpy.html > [2] http://msdn.microsoft.com/en-us/library/vstudio/xdsywd25.aspx This needs tests.

1 0

Fwd: [AppDB] Submitted test data deleted
by Ken Sharp Aug. 5, 2013

Aug. 5, 2013

Deleting test results because the Wine version is old? What the Hell is the plan there? -------- Original Message -------- Subject: [AppDB] Submitted test data deleted Date: Mon, 05 Aug 2013 15:48:43 -0500 From: AppDB <appdb-noreply(a)winehq.org> Reply-To: AppDB <appdb-noreply(a)winehq.org> To: appdb-noreply(a)winehq.org Submitted test data deleted ------------------------------------------------------- The test report you submitted for 'Microsoft Access 2007' has been deleted. The action was performed by Joerg Schiermeier Reasons given Old wine - deprecated. Best regards. The AppDB team http://appdb.winehq.org/ If you don't want to receive any other e-mail, please change your preferences: http://appdb.winehq.org/preferences.php

1 0

wine mannual is wrong?
by 中川祥 Aug. 5, 2013

Aug. 5, 2013

I'm interpretering wine.man. I found it says"this requires X11 to run". But the newest ver. does not require X11,does it?

3 2

Draft: improve IMC/IMCC related functions
by Qian Hong Aug. 5, 2013

Aug. 5, 2013

Hello, This is a follow up of http://www.winehq.org/pipermail/wine-devel/2013-July/100459.html ( [PATCH] imm32: Fixed crashing in ImmGetIMCCSize. ) I believe the attached patchset is the correct fix for the Office crashing bug, how ever, I'm looking for a more graceful way to improve ImmCreateContext and other IMC functions, I appreciate to any feedback on the patches especially Patch 6/6. My tests show that Windows return a HGLOBAL hmem for ImmCreateIMCC() as the returned HIMCC, so it is easy to re-implement ImmCreateIMCC in a graceful way. However, Windows return a magic number for ImmCreateContext, (for instance, have a look at 000B01D9 in the test show in https://testbot.winehq.org/JobDetails.pl?Key=26619&log_207=1#k207) I have no idea how that magic number comes, while I understand there is no need to re-implement ImmCreateContext in the same way as Windows does, I'm still interesting in a more beautiful implementation like my ImmCreateIMCC reached, get rid of casting InputContextData to HIMC if possible. Thanks in advance for any suggestions! If there is no better way, I'll send this series to wine-patches. -- Regards, Qian Hong - http://www.winehq.org

1 0

Re: inputscope.idl: Imported from mingw-w64.
by Ken Sharp Aug. 5, 2013

Aug. 5, 2013

On 05/08/13 12:14, Jacek Caban wrote: > + * No warranty is given; refer to the file DISCLAIMER.PD within this package. It's a minor point but this may be a bit confusing given the file doesn't exist.

1 0

Re: msvcrt: invoke invalid_handler in _wcscpy_s
by Piotr Caban Aug. 5, 2013

Aug. 5, 2013

On 08/02/13 20:08, Daniel Lehman wrote: > + if(!MSVCRT_CHECK_PMT(wcDest)) > + { > + *MSVCRT__errno() = MSVCRT_EINVAL; > return MSVCRT_EINVAL; > + } The MSVCRT_CHECK_PMT macro sets the errno value. There's also MSVCRT_CHECK_PMT_ERR macro if you need to set different error.

1 0