October 2015 - Wine-Devel - list.winehq.org

Re: [PATCH 2/3] msvcp120: implement _Thrd_current

by Piotr Caban

On 10/29/15 16:47, Piotr Caban wrote: > On 10/29/15 15:50, Daniel Lehman wrote: >> +_Thrd_t __cdecl _Thrd_current(void) > The __cdecl calling convention is not compatible in this case. You will > need to do something similar as is done in _Getcoll implementation. > > Unless I'm missing something, quick look on _Getcoll shows that it's > broken when _MSVCP_VER>=110 and 64-bit wine is compiled. I'll take a > look on it. I've fixed _Getcoll implementation today. You may take a look on it now (e.g. thanks to Alexandre change it uses C_ASSERT instead of assert). Thanks, Piotr

Oct. 30, 2015

Re: [PATCH 01/4] msvcp110: Add tr2_sys__Open_dir implementation.(try 4)

by Piotr Caban

On 10/29/15 16:28, YongHao Hu wrote: > + char temppath[MAX_PATH+1]; > + TRACE("(%p %s %p %p)\n", target, debugstr_a(dest), err_code, type); > + if(strlen(dest) >= MAX_PATH) { > + *err_code = ERROR_BAD_PATHNAME; > + return NULL; > + } > + strcpy(temppath, dest); > + strcat(temppath, "\\*"); There may still write outside of temppath array in this code. Thanks, Piotr

Oct. 30, 2015

Re: [1/2] ole32: Accept TYMED_FILE as a possible source of CF_METAFILEPICT format.

by Huw Davies

On 30 Oct 2015, at 06:07, Dmitry Timoshkov wrote: > Huw Davies <huw(a)codeweavers.com> wrote: >> >> This should be easy enough to test. Create a stand-alone data cache >> ( using CreateDataCache() ), then call IOleCache_Cache with the >> relevant fmtetc. > > Ufortunately it's not that simple. IOleCache_Cache() refuses to accept > FORMATETC for CF_METAFILEPICT with medium set to TYMED_FILE, and both > IOleCache_SetData() and IOleObject_SetData() don't accept STGMEDIUM set > to TYMED_FILE either (with properly initialized lpszFileName pointing to > a valid metafile). I have no idea how an existing storage may have medium > set to TYMED_FILE, or how to set it for testing purposes besides manually > patching the storage, any ideas? In that case you'll have to create the stream by hand and try to load it (and also fix your patch so that IOleCache_* don't end up accepting TYMED_FILE). Huw.

Oct. 30, 2015

Re: [PATCH 1/3] msvcp120: implement _Thrd_sleep/yield (try 2)

by Piotr Caban

On 10/29/15 23:04, Daniel Lehman wrote: > + ok((diff > 2000 - TIMEDELTA) && (diff < 2000 + TIMEDELTA), "got %d", diff); This test is occasionally failing on my Windows 7 box (is most cases diff ~= 2165, biggest value I've seen is diff = 2771). I think that it should be enough to just test diff>2000-TIMEDELTA. New line character is also missing here. Thanks, Piotr

Oct. 30, 2015

Re: Microsoft shared PDB fileformat

by Qian Hong

On Fri, Oct 30, 2015 at 6:48 PM, Marcus Meissner <meissner(a)suse.de> wrote: > https://github.com/Microsoft/microsoft-pdb Also this mailing thread in llvm list: RFC: CodeView debug info emission in Clang/LLVM https://groups.google.com/forum/#!topic/llvm-dev/PMqV8Zht5ZM -- Regards, Qian Hong - http://www.winehq.org

Oct. 30, 2015

Microsoft shared PDB fileformat

by Marcus Meissner

Hi, just in from the hell-freezes-over department: https://github.com/Microsoft/microsoft-pdb Ciao, Marcus

Oct. 30, 2015

Re: wldap32: Remove superfluous const in ldap_init declarations.

by Thomas Faber

On 2015-10-19 12:15, Henri Verbeet wrote: > On 19 October 2015 at 11:54, Hans Leidekker <hans(a)codeweavers.com> wrote: >> I'm not sure that's a reason to differ from the psdk though. Note that > I can't answer that for you, but personally I'm usually doing the > reverse. I.e., only matching the PSDK when there's a good reason to. > Usually that means either things that affect the ABI or things that > affect the API in a way that would be visible to winelib applications. Thanks for your reviews! I do see that my patch created an inconsistency; sending a new version now. I'm going with adding the 'const' to the implementation now simply because it's the smaller change, I'm happy to remove it from both headers instead if that's the general (or Alexandre's) preference. Thanks. -Thomas

Oct. 30, 2015

Re: d3d8types.h: Convert the D3DVSD_...SHIFT constants to unsigned.

by Gerald Pfeifer

On Mon, 26 Oct 2015, Henri Verbeet wrote: > I don't think it really makes sense to make the shifts unsigned. There > might be an argument for consistently using unsigned constants unless > they need to be signed, but that would be a much larger change. Okay. If you want I could have a look, say converting one header file, but I agree, it's not much of a priority. Let me know in case. Gerald

Oct. 30, 2015

Re: [PATCH] xmllite/tests: Fix memory leak (valgrind)

by Nikolay Sivov

Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com>

Oct. 30, 2015

Re: [PATCH] msctf/tests: Fix memory leaks (valgrind)

by Marvin

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 https://testbot.winehq.org/JobDetails.pl?Key=17842 Your paranoid android. === w1064 (32 bit inputprocessor) === inputprocessor.c:791: Test failed: Deactivation came unexpectedly

Oct. 30, 2015