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

Re: Norwegian language files using wrong locale

by Ove Kaaven

Daniel Nylander skrev: > On ons, 2008-01-02 at 18:29 +0800, Dmitry Timoshkov wrote: >> "Daniel Nylander" <po(a)danielnylander.se> wrote: >> >>> I noticed that the Norwegian (bokmål) locale files are using the locale >>> "no". Correct locale for Norwegian Bokmål is "nb" (Nynorsk is "nn"). >> I don't see a lower cased "no" in dlls/kernel32/nls/nor.nls. Where did you >> notice that? Wine has the same data as for instance my XP SP2 reports: >> LOCALE_SISO3166CTRYNAME "NO" >> LOCALE_SISO639LANGNAME "nb" > > I did see them in the git tree. The files include "LANGUAGE > LANG_NORWEGIAN, SUBLANG_NORWEGIAN_BOKMAL" (which is correct) but what > happens when Norwegian Nynorsk (no_NN) is added? You've got it reversed. The proper locale setting is nn_NO. The Nynorsk is in non.nls.

Jan. 2, 2008

Re: [5/5] wined3d: change colorkey fixup for stage 0 alphaop / alphaarg

by Stefan Dösinger

Am Mittwoch, 2. Januar 2008 12:44:24 schrieb Stefan Dösinger: > Am Mittwoch, 2. Januar 2008 13:38:57 schrieb Alexander Dorofeyev: > > In case of colorkey emulation for a texture, if the application wants to > > select alpha from diffuse color at stage 0, a fixup to modulate with > > texture alpha more closely matches what the application wants than fixup > > to just select alpha from texture. I tested pop3d and Moto Racer 2. Pop3d still works as expected, but moto racer 2 has some missing geometry. It draws some opaque parts of the scene with color keying on, and a diffuse alpha of 0.0, so the alpha test kicks in and the geometry is missing. The moto racer 2 demo is available online, e.g. http://downloads.gamezone.com/demos/d1576.htm. Note that the installer refuses to run, unless you set your X server's depth to 16. I also caused a regression in that game when I fixed some bugs with native d3drm.dll, so you need the attached hack. (I didn't have time and energy to write a test to test that, but I think that BRANCHFORWARD tokens with offset 0 are special, and that the existing executebuffer test in d3d.c just points out the special case, but the common case is implemented incorrectly)

Jan. 2, 2008

Re: Pixel Shader Problem in 0.9.52

by Stefan Dösinger

Am Mittwoch, 2. Januar 2008 00:04:30 schrieb László Tari: > Hi, > > Problem occurs when Depth Of Filed is turned on in Tomb Raider Anniversary. > > I simply overwrote the 0.9.52 wined3d.dll.so file with the 0.9.44 > wined3d.dll.so in /usr/lib/wine, it works, but a little bit slower. This doesn't look like a pixel shader problem, more like a blitting or offscreen rendering problem. Can you run a regression test and file a bug?

Jan. 2, 2008

Re: #4 msvcrt: Correct search strategy for the wide exec/spawn functions.

by Alexandre Julliard

Hans Leidekker <hans(a)it.vu.nl> writes: > Changelog > Correct search strategy for the wide exec/spawn functions. I don't think we want to duplicate all that code. The Ascii functions should convert their arguments and call the Unicode ones. -- Alexandre Julliard julliard(a)winehq.org

Jan. 2, 2008

Re: include: added the new Vista theme schema headers.

by Alexandre Julliard

"Reece Dunn" <msclrhd(a)googlemail.com> writes: > diff --git a/include/tmschema.h b/include/tmschema.h > index 910de19..3a3e3e3 100644 > --- a/include/tmschema.h > +++ b/include/tmschema.h > @@ -20,6 +20,10 @@ > #ifndef __WINE_TMSCHEMA_H > #define __WINE_TMSCHEMA_H > > +#if defined(__WINE_VSSYM32_H) > +# error tmschema.h (XP) cannot be included along side vssym32.h (Vista). > +#endif That's not the way it's done in the SDK. -- Alexandre Julliard julliard(a)winehq.org

Jan. 2, 2008

Re: Norwegian language files using wrong locale

by Alexandre Julliard

Daniel Nylander <po(a)danielnylander.se> writes: > I did see them in the git tree. The files include "LANGUAGE > LANG_NORWEGIAN, SUBLANG_NORWEGIAN_BOKMAL" (which is correct) but what > happens when Norwegian Nynorsk (no_NN) is added? The filenames should be > changed according to ISO 639-1. But I could be wrong. There's no reason to use a separate file for every variant. Both Norwegian locales should go into the same No.rc file. -- Alexandre Julliard julliard(a)winehq.org

Jan. 2, 2008

Re: [2/5][try 2] ddraw/tests: add test for rendering vertices with zero rhw

by Alexandre Julliard

Alexander Dorofeyev <alexd4(a)inbox.lv> writes: > Fixed some issues pointed out by Stefan Dösinger. > --- > dlls/ddraw/tests/visual.c | 41 +++++++++++++++++++++++++++++++++++++++++ > 1 files changed, 41 insertions(+), 0 deletions(-) This is failing occasionally here: ../../../tools/runtest -q -P wine -M ddraw.dll -T ../../.. -p ddraw_test.exe.so visual.c && touch visual.ok visual.c:797: Test failed: Got color 00000000, expected 00ffffff make: *** [visual.ok] Error 1 -- Alexandre Julliard julliard(a)winehq.org

Jan. 2, 2008

Re: [5/5] wined3d: change colorkey fixup for stage 0 alphaop / alphaarg

by Stefan Dösinger

Am Mittwoch, 2. Januar 2008 13:38:57 schrieb Alexander Dorofeyev: > In case of colorkey emulation for a texture, if the application wants to > select alpha from diffuse color at stage 0, a fixup to modulate with > texture alpha more closely matches what the application wants than fixup to > just select alpha from texture. I am not sure about this. From my understanding of color keying, the color keying should only depend on the texture, not the diffuse color. Unfortunately, this is mutually exclusive with always honoring the app's blending settings. I have two games which use color keying, Moto Racer 2 and Prince of Persia 3D. I'll test the patches with those two games. The problem is that our color keying is emulated, so we'll always have to fiddle to get it working properly everywhere. The correct, but complex and incomplete fix would be a replacement shader for the fixed function pipeline which can implement color keying properly using 'KIL' or 'discard'. Unfortunately that won't fix the problem for cards without fragment shader support. The other patches look good to me

Jan. 2, 2008

Re: [3/5] ddraw/tests: add test for D3DRENDERSTATE_TEXTUREMAPBLEND state

by Stefan Dösinger

Am Mittwoch, 2. Januar 2008 09:21:14 schrieb H. Verbeet: > On 02/01/2008, Alexander Dorofeyev <alexd4(a)inbox.lv> wrote: > > Tests of behavior introduced by patches 4/5 and 5/5. > > --- > > Going by the commit message, I assume this test would currently fail? > You can't do that. You either have to send the patches that fix the > problem first, or you have to wrap the failing tests in todo_wine > blocks, and remove the todo_wine blocks in the patch that fixes the > tests they wrap. Appart of this issue I recommend to move the D3D1 creation into a separate function. Eventually more D3D1 tests will be written, and we want to discourage copypasting that huge block of code.

Jan. 2, 2008

RE: dlls/wintab32/context.c -- simpflication and HEADS UP!

by Rolf Kalbermatter

Gerald Pfeifer wrote on Tuesday, January 01, 2008 6:03 PM: This code obviously was written with signed operands in mind. I wonder if they could get signed somewhere and this function simply uses the wrong datatype. > (I did not get _any_ response for my first mail a month ago > and I still believe the else-part looks bogus, but can we get > at least this first cleanup in?) In the else statement InExt could be 0 (either that or OutExt would be 0) which would cause a division by 0 exception. Definitely doesn't sound right to me. But I have no idea what that code is supposed to do. > @@ -160,10 +160,10 @@ int TABLET_PostTabletMessage(LPOPENCONTE > static inline DWORD ScaleForContext(DWORD In, DWORD InOrg, > DWORD InExt, DWORD > OutOrg, DWORD OutExt) { > - if (((InExt > 0 )&&(OutExt > 0)) || ((InExt<0) && (OutExt < 0))) > - return ((In - InOrg) * abs(OutExt) / abs(InExt)) + OutOrg; > + if ((InExt > 0) && (OutExt > 0)) > + return ( (In - InOrg) * OutExt / InExt ) + OutOrg; > else > - return ((abs(InExt) - (In - InOrg))*abs(OutExt) / > abs(InExt)) + OutOrg; > + return ( (InExt - (In - InOrg)) * OutExt / InExt ) + OutOrg; > } > > LPOPENCONTEXT AddPacketToContextQueue(LPWTPACKET packet, HWND hwnd) Rolf Kalbermatter

Jan. 2, 2008