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

Re: Nightly performance regression graphs
by Dan Kegel 18 May '10

18 May '10

On Tue, May 18, 2010 at 1:13 AM, Sven Baars <sven.wine(a)gmail.com> wrote: >> http://kegel.com/wine/yagmarkdata/wine-1.1.44-245.html > > Just a question. What's this about? > > Vista Wine ratio > heaven2_d3d9_Video_memory 1024.00 256.00 0.25 The video card definitely has a gigabyte of video memory, and I just confirmed that the wineprefix running the test has user.reg:"VideoMemorySize"="1024" so I have no idea. Sounds like a bug, but I doubt it affects the results. A little quality time with +relay would probably show what's going on. (Or Wine could buy a copy of the source code for this benchmark. We might want one anyway...) - Dan

1 0

Re: [PATCH 1/4] wined3d: only activate blit_shader in arbfp_blit_surface for source fixups
by Henri Verbeet 18 May '10

18 May '10

On 18 May 2010 09:30, Roderick Colenbrander <thunderbird2k(a)gmail.com> wrote: > - arbfp_blit_set((IWineD3DDevice *)device, src_surface); > + if (is_complex_fixup(src_surface->resource.format_desc->color_fixup) && > + !is_complex_fixup(dst_surface->resource.format_desc->color_fixup)) > + arbfp_blit_set((IWineD3DDevice *)device, src_surface); > - This doesn't do what your subject line says it does. - Why do you think this patch makes sense?

2 4

Re: [PATCH 3/4] wined3d: allow complex -> complex blits in arbfp blit_shader under special circumstances
by Henri Verbeet 18 May '10

18 May '10

On 18 May 2010 09:30, Roderick Colenbrander <thunderbird2k(a)gmail.com> wrote: > This patch should also address bug 22575 when using the arbfp backend. A similar fix is needed > for the ffp backend. That's meaningless as a Changelog entry. > + && (dst_usage & WINED3DUSAGE_RENDERTARGET)) What is that condition for?

2 1

Re: advapi32: Remove redundant "is NULL and error out" code (coccicheck).
by Michael Stefaniuc 18 May '10

18 May '10

Alexandre, please disregard this patch as I missed somehow that the error codes are different. And to figure out which to use needs a test. bye michael Michael Stefaniuc wrote: > --- > dlls/advapi32/crypt.c | 5 ----- > 1 files changed, 0 insertions(+), 5 deletions(-) > > diff --git a/dlls/advapi32/crypt.c b/dlls/advapi32/crypt.c > index bb019d0..395879f 100644 > --- a/dlls/advapi32/crypt.c > +++ b/dlls/advapi32/crypt.c > @@ -1413,11 +1413,6 @@ BOOL WINAPI CryptGenKey (HCRYPTPROV hProv, ALG_ID Algid, DWORD dwFlags, HCRYPTKE > > TRACE("(0x%lx, %d, %08x, %p)\n", hProv, Algid, dwFlags, phKey); > > - if (!prov) > - { > - SetLastError(ERROR_INVALID_HANDLE); > - return FALSE; > - } > if (!phKey || !prov || prov->dwMagic != MAGIC_CRYPTPROV) > { > SetLastError(ERROR_INVALID_PARAMETER);

1 0

Re: [PATCH 4/4] wined3d: remove unneeded memory cleanup code from BltOverride
by Henri Verbeet 18 May '10

18 May '10

On 18 May 2010 09:30, Roderick Colenbrander <thunderbird2k(a)gmail.com> wrote: > We think the code is not needed and the code was wrong anyway since it didn't check whether the pointers it freed > where valid. In which cases can it be invalid?

1 0

Re: [PATCH] ntdll: Use the old mmap() wrapper on 32-bit Mac OS. (try 3)
by Charles Davis 18 May '10

18 May '10

What's wrong with my patch now? It's a really simple fix; in fact, this is as about as simple as it gets. How does this "break libwine compatibility"? Chip

3 7

Re: [PATCH] comctl32/listview: Derive subitem rect from listview origin
by Nikolay Sivov 17 May '10

17 May '10

On 5/18/2010 02:36, Andrew Eikum wrote: > > The old code used the containing item's position to determine the > subitem's rect. Instead, we should be using the entire listview's > origin as the offset, since the item position can change as columns > are inserted and re-arranged, leading to broken listviews. > --- > dlls/comctl32/listview.c | 4 ++-- > 1 files changed, 2 insertions(+), 2 deletions(-) Looks right. Seems to be (one?) of remaining problem that comes out after I added column reordering. Please add a minimal test case for this to prevent a regression. All you need - add 2 columns, add item, reorder columns with a LVM_SETCOLUMNORDERARRAY and test rectangle.

1 0

Re: [PATCH 4/4] wined3d: add BLIT_OP_COMPLEX_UPLOAD
by Henri Verbeet 17 May '10

17 May '10

On 17 May 2010 13:11, Roderick Colenbrander <thunderbird2k(a)gmail.com> wrote: > This patch introduces a new blit operation to allow complex uploads from LoadLocation. It allows hw accelerated > p8 uploads to function again (I broke them by accident). Without this blit_operation there is no way to detect > this single case at the moment in which we allow the P8 shader (remember both the source and destination have > complex fixups set, so this doesn't pass). I'm not so convinced this needs a separate blit operation, it should always be possible to do blits between surfaces with the same formats as long as no stretching is involved.

2 1

Re: urlmon: Added tests for some of the property methods. Added a dependency to urlmon for oleaut32
by Jacek Caban 17 May '10

17 May '10

Hi Thomas, On 5/17/10 3:05 AM, Thomas Mullaly wrote: > Added tests to make sure IUri_GetPropertyBSTR / IUri_GetPropertyDWORD > / IUri_GetPropertyLength handle invalid arguments correctly. Fixed > up/added some comments. > > Had to add a dependency to urlmon for oleaut32. This is because the > Uri interface returns all the string properties of a URI as BSTR's and > it needs the SysAllocString/SysFreeString functions in oleau32. I > checked on windows using Dependency Walker and the native version of > urlmon has this same dependency. -IMPORTS = uuid ole32 rpcrt4 shlwapi wininet user32 advapi32 kernel32 ntdll +IMPORTS = uuid ole32 oleaut32 rpcrt4 shlwapi wininet user32 advapi32 kernel32 ntdll You should add oleaut32 dependency in a patch that uses its functions. hr = IUri_GetPropertyBSTR(uri, i,&str, 0); + todo_wine { + ok(hr == S_OK, "Error: GetPropertyBSTR returned 0x%08x, expected 0x%08x. property=%d\n", + hr, S_OK, i); + } + todo_wine { ok(str != NULL, "Error: Expected the BSTR not to be NULL, property=%d\n", i); } + ok(!SysStringLen(str), "Error: Expected the BSTR to be of length 0, but it was %d instead. property=%d\n", + SysStringLen(str), i); You will have to add tests for the exact values returned sooner or later, so this will be tested better then. There is not much value in tests like this. + if(str) SysFreeString(str); You don't need NULL check here. Thanks, Jacek

1 0

Re: riched20: Avoid using long.
by testbot＠testbot.winehq.org 17 May '10

17 May '10

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=2105 Your paranoid android. === W7PRO (32 bit editor) === editor.c:4696: Test failed: test paste: strcmp = 1, text='testing paste editor.c:4714: Test failed: test paste: strcmp = 1 editor.c:4726: Test failed: test paste: strcmp = -1 editor.c:4756: Test failed: test paste: strcmp = -1, actual = 'Ts' editor.c:5971: Test failed: Cursor is at 0 instead of 4 editor.c:5977: Test failed: Cursor is at 0 instead of 9 editor.c:5983: Test failed: Cursor is at 0 instead of 4 editor.c:5997: Test failed: Cursor is at 8 instead of 9 editor.c:6005: Test failed: Cursor is at 11 instead of 9 editor.c:6016: Test failed: Cursor is at 0 instead of 8 editor.c:6041: Test failed: Cursor is at 0 instead of 8

2 1