October 2017 - Wine-devel - list.winehq.org

Re: [PATCH 3/4] d3d8/tests: Make test_reset more resilient against DPI scaling. (v2)
by Marvin 19 Oct '17

19 Oct '17

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=33639 Your paranoid android. === w1064 (32 bit device) === device.c:1532: Test failed: Reset failed, hr 0x80070057. device.c:1534: Test failed: TestCooperativeLevel failed, hr 0x88760869. device.c:1546: Test failed: D3DVIEWPORT->Width = 400. device.c:1547: Test failed: D3DVIEWPORT->Height = 300. 0864:device: unhandled exception c0000005 at 706EA86E

1 0

Re: [PATCH v2] user32: Add the IDI_SHIELD icon resource.
by Alexandre Julliard 19 Oct '17

19 Oct '17

"Sergio Gómez Del Real" <sergio.g.delreal(a)gmail.com> writes: > Applications such as Fidibo > (https://bugs.winehq.org/show_bug.cgi?id=43770) ask for the IDI_SHIELD > OEM resource, for example by calling comctl32.LoadIconMetric. Because > the resource doesn't exist in user32, the function returns error and, > in the specific case of Fidibo, the application crashes. This patch > adds to user32.rc the appropriate lines and adds the file > oic_shield.svg, from which oic_shield.ico will be generated. > > NOTE: Tango's public domain network-wireless-encrypted.svg file was > used. Thanks to Alex Henrie for this. To generate a proper icon, we'd need multiple scaled images for the various sizes, and bitmaps for low-color resolutions, with the labels expected by the buildimage script. You can check how this is done in oic_winlogo.svg. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH 5/6] wined3d: Pack interpolation modes.
by Henri Verbeet 18 Oct '17

18 Oct '17

On 18 October 2017 at 00:25, Józef Kucia <jkucia(a)codeweavers.com> wrote: > +static inline DWORD wined3d_extract_bits(const DWORD *bitstream, > + unsigned int offset, unsigned int count) > +{ > + const unsigned int word_bit_count = sizeof(*bitstream) * CHAR_BIT; > + const unsigned int idx = offset / word_bit_count; > + const unsigned int shift = offset % word_bit_count; > + DWORD mask = (1u << count) - 1; It's not an issue in this patch, but note that the above doesn't work for count > 31. We probably don't care about count > 32, but we may care about count == 32. Specifically, "1u << 32" is undefined, but in practice tends to evaluate as "1u << 0".

2 3

Re: [PATCH 2/4] d3d9/tests: Fix test_reset on Nvidia Windows 10.
by Stefan Dösinger 18 Oct '17

18 Oct '17

> Am 18.10.2017 um 21:38 schrieb Stefan Dösinger <stefan(a)codeweavers.com>: > > The client rectangle is 204x150 on my system (real one, Nvidia GPU, not > qemu), even though we requested 200x150. Odd, but not really the concern > of this test. I had an epiphany where the odd behavior comes from: It was caused by my custom UI scaling factor of 172% (=165 DPI). Setting it to 150% (144 DPI), a value suggested by Windows, gives the expected window size. I still think the patch is a good idea because it makes the test more robust against such configurations. I’ll resend it tomorrow with an updated title and description.

1 0

Re: [PATCH] user32: Add the IDI_SHIELD icon resource.
by Alex Henrie 18 Oct '17

18 Oct '17

> NOTE: oic_shield.svg is just a copy of oic_winlogo.svg. Could we use an actual shield icon instead? It was pretty easy for me to open Tango's network-wireless-encrypted.svg in Inkscape and edit it down to a plain shield. This icon is in the public domain, so it's fine to use it in Wine. -Alex

1 0

Re: [PATCH] winmm/tests: Give play from 0 to 0 more time to finish.
by Marvin 18 Oct '17

18 Oct '17

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=33612 Your paranoid android. === w864 (64 bit mci) === mci.c:995: Test failed: got 0001 instead of MCI_NOTIFY_xyz 0000 from command play (nowait) mci.c:998: Test failed: mci pause wait returned MCIERR_NONAPPLICABLE_FUNCTION mci.c:1002: Test failed: mci status mode: stopped mci.c:1003: Test failed: got 0001 instead of MCI_NOTIFY_xyz 0002 from command play mci.c:1004: Test failed: Expect message 0001 from status

2 1

Re: [1/2] kerberos: Add a stub for Kerberos5 Authentication Package. (v2)
by Alexandre Julliard 18 Oct '17

18 Oct '17

Dmitry Timoshkov <dmitry(a)baikal.ru> writes: > When compiling without HAVE_KRB5_KRB5_H defined the compiler issues > a warning about libkrb5_handle not being used, libkrb5_handle will be > used later as a detection that libkrb5 is available. I'll move this to avoid the warning, it can be moved back when it's used. Though it may be cleaner to always check it inside the ifdef anyway. -- Alexandre Julliard julliard(a)winehq.org

1 0

wine-2.0.3-rc2 Available
by Michael Stefaniuc 17 Oct '17

17 Oct '17

Hello guys, I have pushed wine-2.0.3-rc3 (with tag) to https://github.com/mstefani/wine-stable Changes since rc1: - Removed the commit "gdi32: Return non-zero VREFRESH value for display devices." aka bug #43369 It is a no-op without cherry-picking some risky other patches. bye michael

1 0

Re: [PATCH 3/5] gdiplus: Add internal string format properties used by typographical format
by Vincent Povirk 17 Oct '17

17 Oct '17

This removes margins when a NULL format is specified. If that change is correct, it should probably be in a separate patch and have a test case.

2 1

Re: [1/2] kerberos: Add a stub for Kerberos5 Authentication Package.
by Hans Leidekker 17 Oct '17

17 Oct '17

Hi Dmitry, Thanks. Alexandre pointed out that we do not need to load libkrb5 dynamically because libgssapi_krb5 depends on libkrb5. So if libkrb5 is missing the SSP wouldn't be usable anyway. > diff --git a/configure.ac b/configure.ac > index 58737162a8..82faa4b943 100644 > --- a/configure.ac > +++ b/configure.ac > @@ -422,6 +422,7 @@ AC_CHECK_HEADERS(\ > ieeefp.h \ > inet/mib2.h \ > io.h \ > + krb5/krb5.h \ > kstat.h \ > libproc.h \ > link.h \ > @@ -1625,6 +1626,14 @@ fi > WINE_NOTICE_WITH(gsm,[test "x$ac_cv_lib_soname_gsm" = "x"], > [libgsm ${notice_platform}development files not found, gsm 06.10 codec won't be supported.]) > > +dnl **** Check for krb5 **** > +if test "$ac_cv_header_krb5_krb5_h" = "yes" > +then > + WINE_CHECK_SONAME(krb5,krb5_init_context) > +fi > +WINE_NOTICE_WITH(krb5,[test "x$ac_cv_lib_soname_krb5" = "x"], > + [libkrb5 ${notice_platform}development files not found, krb5 won't be supported.]) > + There is pkgconfig support for krb5. I guess we want to use it. You need to guard the code that uses krb5 functions with an ifdef or it will fail to compile if the headers are missing.

4 5