November 2017 - Wine-Devel - list.winehq.org

Re: [PATCH] d3dx9/tests: Relax D3DXQuaternionToAxisAngle test stringency a bit
by Matteo Bruni Nov. 10, 2017

Nov. 10, 2017

Signed-off-by: Matteo Bruni <mbruni(a)codeweavers.com>

1 0

Switch to wine-devel for patch submissions
by Alexandre Julliard Nov. 10, 2017

Nov. 10, 2017

Folks, The wine-devel list can now be used for submitting patches. Please start using it instead of wine-patches for all patch submissions and signoff replies. For the time being, both lists feed into the patch tracker. Once we have confirmed that everything works, wine-patches will be retired. The tracker should be able to figure out which mails are patches, but for the benefit of human readers, please add a [PATCH] notice in the subject line of your mails. Please let me know if you notice anything broken. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: [PATCH] d3dx9/tests: Use a macro to initialize matrix rows in math tests
by Henri Verbeet Nov. 10, 2017

Nov. 10, 2017

On 10 November 2017 at 08:06, Alex Henrie <alexhenrie24(a)gmail.com> wrote: > diff --git a/dlls/d3dx9_36/tests/math.c b/dlls/d3dx9_36/tests/math.c > index ec07348685..409e3cb650 100644 > --- a/dlls/d3dx9_36/tests/math.c > +++ b/dlls/d3dx9_36/tests/math.c > @@ -25,6 +25,8 @@ > > #define ARRAY_SIZE(array) (sizeof(array) / sizeof((array)[0])) > > +#define SET_ARRAY4(array, a, b, c, d) array[0] = a; array[1] = b; array[2] = c; array[3] = d You don't need a macro for that.

2 2

Re: [Resend PATCH] kernel32: Handle zh-Hant and zh-Hans in LocaleNameToLCID.
by Alexandre Julliard Nov. 10, 2017

Nov. 10, 2017

Zhiyi Zhang <yi.gd.cn(a)gmail.com> writes: > Superseded patch 138395. > > LocaleNameToLCID now can handle zh-Hant and zh-Hans correctly with > get_default_children().get_default_children() acts as a map from a > parent language to its default child language.Since zh-Hant and > zh-Hans are not actual locale identifiers. All locale info of zh-Hant > and zh-Hans are directly mapped to its own child language, zh-HK and zh-CN. The LOCALE_SPARENT data is a good addition, but this could be split to a separate patch. Also you should be able to extend the existing get_default_sublang() logic instead of adding another table of defaults. -- Alexandre Julliard julliard(a)winehq.org

2 2

[PATCH] kernel32/nls: Add LOCALE_SPARENT entries.
by Zhiyi Zhang Nov. 10, 2017

Nov. 10, 2017

1 0

Re: [PATCH v2] d3dx9: Reduce rounding error in D3DXQuaternionToAxisAngle
by Henri Verbeet Nov. 9, 2017

Nov. 9, 2017

On 9 November 2017 at 20:04, Alex Henrie <alexhenrie24(a)gmail.com> wrote: > v2: Change 2.0f to simply 2.0 to make it clear that truncation is done > after multiplication. Mostly for what it's worth, note that that makes no difference for a multiplication by two. The exponent is incremented, the significand stays the same.

2 1

Re: [PATCH] d3dx10: Add D3DX10PreprocessShaderFromMemory stub.
by Henri Verbeet Nov. 9, 2017

Nov. 9, 2017

On 9 November 2017 at 17:55, Andrey Gusev <andrey.goosev(a)gmail.com> wrote: > +HRESULT WINAPI D3DX10PreprocessShaderFromMemory(const char *data, SIZE_T data_size, const char *filename, > + const D3D10_SHADER_MACRO *defines, ID3DInclude *include, ID3DX10ThreadPump *pump, ID3D10Blob **shader_text, > + ID3D10Blob **errors, HRESULT *hresult) > +{ > + FIXME("data %s, data_size %lu, filename %s, defines %p, include %p, pump %p, shader_text %p, errors %p, " > + "hresult %p stub!\n", > + debugstr_a(data), data_size, debugstr_a(filename), defines, include, pump, shader_text, errors, hresult); It's questionable how useful debugstr_a() is for "data", but it's also wrong. "data" is not necessarily \0 terminated, "data_size" specifies its length. > -@ stub D3DX10PreprocessShaderFromMemory(str long str ptr ptr ptr ptr ptr) > +@ stdcall D3DX10PreprocessShaderFromMemory(str long str ptr ptr ptr ptr ptr ptr) Likewise.

2 1

Re: d3dx9_36: Silence an uninitialized compiler warning (GCC 8).
by Paul Gofman Nov. 9, 2017

Nov. 9, 2017

There is actually no usage of uninitialized variable there, though it is probably not exactly straightforward for compiler to guess. Why it does not complain that 'cdesc[index]' may be used uninitialized while it is handled the same way? AFAIK GCC 8 is not released yet, maybe this will change before final release? If still to change that, does the attached patch fixes the warning as well (sorry, I don't have gcc 8 in hands to test)? If yes, I would suggest that would be a tiny bit nicer way to silence the warning. On 11/07/2017 01:22 AM, Gerald Pfeifer wrote: > There are more of this in d3dx9_36, but at least this is one down. > > Gerald > > > Signed-off-by: Gerald Pfeifer <gerald(a)pfeifer.com> > --- > dlls/d3dx9_36/preshader.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/dlls/d3dx9_36/preshader.c b/dlls/d3dx9_36/preshader.c > index e3e91c643b..ed22dc4045 100644 > --- a/dlls/d3dx9_36/preshader.c > +++ b/dlls/d3dx9_36/preshader.c > @@ -888,7 +888,7 @@ static HRESULT get_constants_desc(unsigned int *byte_code, struct d3dx_const_tab > for (i = 0; i < desc.Constants; ++i) > { > unsigned int index = out->input_count; > - WORD constantinfo_reserved; > + WORD constantinfo_reserved = 0; > > hc = ID3DXConstantTable_GetConstant(ctab, NULL, i); > if (!hc)

3 5

Re: [PATCH v2] d3dx9: Reduce rounding error in D3DXQuaternionToAxisAngle
by Matteo Bruni Nov. 9, 2017

Nov. 9, 2017

2017-11-09 17:34 GMT+01:00 Alex Henrie <alexhenrie24(a)gmail.com>: > Signed-off-by: Alex Henrie <alexhenrie24(a)gmail.com> > --- > Fixes test failure on 64-bit Linux. > > v2: Change 2.0f to simply 2.0 to make it clear that truncation is done > after multiplication. > --- > dlls/d3dx9_36/math.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/dlls/d3dx9_36/math.c b/dlls/d3dx9_36/math.c > index 8f5f5b2413..1b4edf82cf 100644 > --- a/dlls/d3dx9_36/math.c > +++ b/dlls/d3dx9_36/math.c > @@ -1651,7 +1651,7 @@ void WINAPI D3DXQuaternionToAxisAngle(const D3DXQUATERNION *pq, D3DXVECTOR3 *pax > paxis->z = pq->z; > } > if (pangle) > - *pangle = 2.0f * acosf(pq->w); > + *pangle = 2.0 * acos(pq->w); /* acosf has too much rounding error */ > } > > /*_________________D3DXVec2_____________________*/ > -- > 2.14.2 I think just relaxing the test a tiny bit, allowing 1 ULP of error margin, is preferable in this case.

1 0

Re: [PATCH] d3dx9/tests: Provide 8 significant digits in matrix reference values
by Henri Verbeet Nov. 9, 2017

Nov. 9, 2017

Thanks for fixing these. 2017-11-09 6:10 GMT+01:00 Alex Henrie <alexhenrie24(a)gmail.com>: > Signed-off-by: Alex Henrie <alexhenrie24(a)gmail.com> > --- > Fixes test failures on 64-bit Linux. > > A 32-bit float can hold 7.2 decimal digits, so it's best to provide 8 to > make sure that the correct values are accurately represented. > If it were up to me, I'd prefer these to be in the same %.8e format the failure messages use. > The correct outputs of functions that take simple floats as inputs were > computed by feeding equations into Wolfram Alpha. I think that's probably fine, but note that we don't care about the "correct" values so much as "whatever Windows produces". Hopefully those would be reasonably close, but they're by no means guaranteed to be the same.

3 2