January 2014 - Wine-Devel - list.winehq.org

Re: Command Stream Update
by Forest Jan. 12, 2014

Jan. 12, 2014

On Tue, 07 Jan 2014 22:31:24 +0100, Stefan Dösinger wrote: >I have published an update to the command stream patches from >September. The code is available on Github at >https://github.com/stefand/wine Thanks, Stefan! I used a snapshot of your repository for my latest Ubuntu PPA release. This ought to please the users who have been asking for an update. (Last time I checked, Launchpad said there were a hundred or so people using my builds.) https://launchpad.net/~foresto/+archive/winepatched/ I was pleasantly surprised to find that a long-standing Guild Wars visual bug (reproducible by changing between armor and town clothes) seems to be fixed by your latest code. Here's the bug report: http://bugs.winehq.org/show_bug.cgi?id=29872

1 0

Fwd: nearly all applications immediately crash
by MadLoisae＠gmx.net Jan. 11, 2014

Jan. 11, 2014

Hello wine-users, have an issue with my wine. My system is debian wheezy x86_64 with installed multiarch and wine-bin:i386, the packages are all from debian so it's wine version 1.4.1. nearly all applications crash immediately on useraction. the attached output was from most recent notepad++ installer - I can start it and click on next, I can see the license a very short moment and the installer disappears with an error in console I started from. wine: Unhandled page fault on write access to 0x000000f9 at address 0x7bc4aa6b (thread 0009), starting debugger... err:seh:raise_exception Unhandled exception code c0000005 flags 0 addr 0x7bc4aa6b the attached output is with WINEDEBUG=+relay if I copy a windows-installed Notepad++ to program directory and start it from there, I can open files and immediately it crashes. Sometimes the text is displayed, if I then scroll or do anything in program it ... crashes. :-/ I tried so far some gdi based winetricks, also I tried several times purging wine complete (including removing .wine), did not help. my system is headless, no monitor attached. vncserver runs on it, icewm as window manager. any ideas? Thanks! Alois

1 0

Re: iphlpapi: Return error for IP address with value zero in IcmpSendEcho
by Bruno Jesus Jan. 11, 2014

Jan. 11, 2014

Hi, Alexandre. I see this patch is pending and unfortunately it's not obvious to me what the reason is. The IP 0.0.0.0 need to be stopped otherwise linux will turn it into 127.0.0.1 and successfully ping it. Best wishes, Bruno On Thu, Jan 9, 2014 at 8:45 PM, Bruno Jesus <00cpxxx(a)gmail.com> wrote: > I missed the function name in the subject. Please ignore 101595. > > Fixes http://bugs.winehq.org/show_bug.cgi?id=33272

1 0

Command Stream Update
by Stefan Dösinger Jan. 10, 2014

Jan. 10, 2014

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I have published an update to the command stream patches from September. The code is available on Github at https://github.com/stefand/wine (direct git url at https://github.com/stefand/wine) A small change summary: *) Tons of bug fixes. *) Dynamic surface maps. This improves World of Warcraft and many other games. *) The code avoids acquiring a GL context on the main thread, with few exceptions. This improves stability on all drivers, especially on fglrx. *) This version essentially matches the command stream code we shipped in CrossOver 13, just applied on top of Wine. *) The Wine version this tree is based on is the git head from this evening, 460991a8. As with the last version you have to set HKEY_CURRENT_USER/Software/Wine/Direct3D/CSMT="enabled" to use it. You should disable StrictDrawOrdering because it is no longer needed with CSMT on. I plan to push updates to this repository at irregular intervals. I probably won't send an announcement to wine-devel every time I do this. I also don't promise any schedule - it mostly depends on how busy I am with other work and how difficult the rebases are. Enjoy, Stefan -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJSzHIsAAoJEN0/YqbEcdMw10wQAIBZ1LuNK4By0WLZzujwww3Z mZjd/kgux13THVtTzVoxVL38nzhOxIuGV0B5P6yoXJlKpxILSCdjr208EWMf+qo7 j4m3DGibrSrjPOCog4hXQeMrn2YjHC2BnFtDIiCYKq33ZPjJ60mSKEkpOzzbeEa8 7C7l9Y+mGKHQl6WViUp7HMoxFo3irZ1DCxRiDiOcFXdQhOZEpuuClB+vabUh15OH 3iNQdAwIiH9YZk9ir7nF5zcbnnLJGNjKNNUPEPu91P1PQ9LdM3zNqz7jU2Klo0Kl 7ZA+O6JVHSuWw04w1Lz6UM7Exu1dY6hRL9BfZQxodBOXtC0NwqIMbeW0bKn1VMEL cF4a+3sqGTJhni0dITEP1BzqmZH3sqYBPuOCb244+UOYwUjLTtx2QcASE3khb8D8 XVry6zrGJl1t1WzwPC/SK6/IOWHhvcznX+JeoYQiNIfOuFUsoVjKcswDxDLLi/+0 qUSpdWvI9fWy/udpaJoemspX0fjvNud6TcdLA/hHvfpoa6+fbHEJV+EC0M6PHFw8 RZ6Hi3VTlg0FBKYqowarqjjYNAxYWecTo9LKjMYTzFVpH4fTKcZUOpMdV+A2sRvv F0Hy0beXyih+Q03HwKmIbWQqWME1Fr62iCa88uO6leT4tPJ5SG3oZ2Reb2zmvFJq QbfNO3JMTqTqrXyUcsQj =dysQ -----END PGP SIGNATURE-----

3 8

Re: [website] Make https for wiki
by Jeremy Newman Jan. 10, 2014

Jan. 10, 2014

We cannot apply this patch until the Wiki has a certificate and has been setup with an SSL vhost. -N On 01/10/2014 08:10 AM, Andrey Gusev wrote: > This should fix bug http://bugs.winehq.org/show_bug.cgi?id=34443

1 0

Re: ws2_32/tests: Add a trailing '\n' to a win_skip() call.
by Marvin Jan. 10, 2014

Jan. 10, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4157 Your paranoid android. === wxppro (32 bit sock) === sock.c:557: Test failed: oob_server (4bc): unexpectedly at the OOB mark: 0 === w2008s64 (32 bit sock) === sock.c:557: Test failed: oob_server (750): unexpectedly at the OOB mark: 0 sock: unhandled exception c0000005 at 00000000 === w2008s64 (64 bit sock) === sock.c:557: Test failed: oob_server (654): unexpectedly at the OOB mark: 0

1 0

Re: [2/2] dpnet: Partial Implement IDirectPlay8Peer SetPeerInfo (try 2)
by Bruno Jesus Jan. 10, 2014

Jan. 10, 2014

On Fri, Jan 10, 2014 at 5:38 AM, Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> wrote: > Hi, > Remove pointer checks > > Changelog: > dpnet: Partial Implement IDirectPlay8Peer SetPeerInfo > Hi, I'm not used to review patches but please take a look at the points below. + if(This->peername) + HeapFree(GetProcessHeap(), 0, This->peername); + if(This->data) + HeapFree(GetProcessHeap(), 0, This->data); + You don't need to check for NULL. + len = lstrlenW(pdpnPlayerInfo->pwszName); + This->peername = HeapAlloc(GetProcessHeap(), 0, len * sizeof(WCHAR)); + memcpy(This->peername, pdpnPlayerInfo->pwszName, len * sizeof(WCHAR)); If peername is a string don't you need a len + 1 to mark NULL in the end? I'm not sure but since you can use lstrlenW can't you use heap_strdupW? + This->data = HeapAlloc(GetProcessHeap(), 0, pdpnPlayerInfo->dwDataSize); + memcpy(This->data, pdpnPlayerInfo->pvData, pdpnPlayerInfo->dwDataSize); Don't you need to store the dwDataSize in datasize you created in the struct? > Best Regards > Alistair Leslie-Hughes > Best regards, Bruno

1 0

Re: [2/2] dpnet: Partial Implement IDirectPlay8Peer SetPeerInfo
by Nikolay Sivov Jan. 10, 2014

Jan. 10, 2014

On 1/10/2014 11:18, Alistair Leslie-Hughes wrote: > Hi, > > > Changelog: > dpnet: Partial Implement IDirectPlay8Peer SetPeerInfo > > > Best Regards > Alistair Leslie-Hughes > > + if(This->peername) > + HeapFree(GetProcessHeap(), 0, This->peername); > + if(This->data) > + HeapFree(GetProcessHeap(), 0, This->data); > + You don't need to check for NULL.

1 0

Re: [PATCH 1/2] ddraw: Make DirectDrawEnumerateEx only enumerate secondary display devices when requested.
by Henri Verbeet Jan. 9, 2014

Jan. 9, 2014

I think this looks ok in principle, but a test wouldn't hurt.

2 3

Re: Assorted spelling fixes
by Marvin Jan. 9, 2014

Jan. 9, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4101 Your paranoid android. === wxppro (32 bit domdoc) === domdoc.c:9860: Test failed: got 00244F94

1 0

Re: Command Stream Update
by Yuri Shishenko Jan. 9, 2014

Jan. 9, 2014

Civilization V has performance regression with this wine version. It shows less FPS than unpatched wine-1.7.10: 20 FPS on maximum camera distance while with unpatched wine it shows 40-50 FPS. My PC specs: openSUSE 13.1, linux 3.12.6 x86_64, nVidia GTX660 with 331.20 driver, Intel Core I5-3570k.

1 0

Re: [PATCH 4/5] d3dx9_36/tests: Use a window with a non-zero size.
by Henri Verbeet Jan. 9, 2014

Jan. 9, 2014

On 9 January 2014 13:43, Stefan Dösinger <stefan(a)codeweavers.com> wrote: > WS_SYSMENU | WS_POPUP follows the existing code in D3DXCreateBoxTest, > which is the only test that created a useable device on my system. I think you might as well just use WS_OVERLAPPEDWINDOW like e.g. the d3d9 tests do, but I don't care that much.

1 0

Re: [PATCH 6/6] ddraw/tests: Add some color key tests for complex surfaces.
by Stefan Dösinger Jan. 9, 2014

Jan. 9, 2014

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 2014-01-09 12:17, schrieb Henri Verbeet: > + hr = IDirectDrawSurface7_GetAttachedSurface(surface, &caps, > &tmp); + ok(SUCCEEDED(hr), "Failed to get attached surface, hr > %#x.\n", hr); + hr = IDirectDrawSurface7_GetColorKey(tmp, > DDCKEY_SRCBLT, &color_key); + ok(hr == DDERR_NOCOLORKEY, "Got > unexpected hr %#x, i %u.\n", hr, i); + + > color_key.dwColorSpaceLowValue = 0x0000ff00; + > color_key.dwColorSpaceHighValue = 0x0000ff00; + hr = > IDirectDrawSurface7_SetColorKey(surface, DDCKEY_SRCBLT, > &color_key); + ok(SUCCEEDED(hr), "Failed to set color key, hr > %#x.\n", hr); + memset(&color_key, 0, sizeof(color_key)); + > hr = IDirectDrawSurface7_GetColorKey(surface, DDCKEY_SRCBLT, > &color_key); + ok(SUCCEEDED(hr), "Failed to get color key, hr > %#x.\n", hr); + ok(color_key.dwColorSpaceLowValue == 0x0000ff00, > "Got unexpected value 0x%08x.\n", + > color_key.dwColorSpaceLowValue); + > ok(color_key.dwColorSpaceHighValue == 0x0000ff00, "Got unexpected > value 0x%08x.\n", + color_key.dwColorSpaceHighValue); + > + IDirectDrawSurface_Release(tmp); Did you mean to call SetColorKey and GetColorKey on tmp instead of surface here? For the sake of completion it might be interesting to test SetColorKey on cube textures. 5 of the faces are attached to a root surface but aren't mipmap sublevels. (I guess it'll still return DDERR_NOTONMIPMAPSUBLEVEL.) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJSzpDjAAoJEN0/YqbEcdMwyl0P/3OHO0kK93dh6XdY1RU9IvNv BASe8YHhv1a6rRxYNGDzGoX5bsnefWLcqUnsWsGGjlxdA13BxAB4NCenaMecu49v d3egCH547Bd94K/9wqqlDxl+aEe0Hh86Ig9t0LNBuruNh1hNKTG9OdohzJBuX1VD TcVYRv6S8wEwM2Gb24RL8wN359497vn58G3fAVxK7PiK/edqxlDTg+6oi6P0/L+x HLGFlSVL9bgtxpwAUVh3i9SRsnA6iS07qPmSHOXdoCz/OhoY29Lx7HLFRCOgUbJS XAo00HKWP+qB4myqt3Iir9QgtiFLFZiyBBn0HQQEWGeRsDpZTazH/EMZrbUPhP03 f9AiW/r7XX5fOoRG5HBv7i/eXdOw6XexVLEyXb57sEilNJfBF3ux2wDw/E34AiGT wUTkgY2VwjgaxNl022qsFYz6fboORwbQm0VyvWzxbKw7326pdt6Y3Q2WTvPEQ5Rk JvlVJ8McRzkrYU5hXjRW1D7/2rVmV8z+xGGi6IU1JoS3PBWIYMKjUDRlFg0n+JQG UzLnJcnxm/js8Ah4R8MWje1v4cq3TlKnc7WrMIwNtu3Wmvwjo34s4A84xdVPneY/ jIqgUGFBOeH/XnK+FAdU+ZFuUn4k/ttVNOuSrPe5HsZxc1uxGnec5OtuZB8FTjFH EH0rn+AxEiyaGPON+4w6 =GtY2 -----END PGP SIGNATURE-----

2 2

Re: [PATCH 5/6] ddraw: Return DDERR_NOTONMIPMAPSUBLEVEL for ddraw_surface7_SetColorKey() on mipmap sub levels.
by Marvin Jan. 9, 2014

Jan. 9, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4096 Your paranoid android. === wxppro (32 bit d3d) === d3d.c:1261: Test failed: Homogeneous output was generated despite UNCLIPPED flag === wvista (32 bit d3d) === d3d.c:1261: Test failed: Homogeneous output was generated despite UNCLIPPED flag === w2008s64 (32 bit d3d) === d3d.c:1261: Test failed: Homogeneous output was generated despite UNCLIPPED flag === w7pro64 (32 bit d3d) === d3d.c:1261: Test failed: Homogeneous output was generated despite UNCLIPPED flag

1 0

Re: [PATCH 4/5] devenum: Make CLSID_VideoCompressorCategory a special category.
by Alexandre Julliard Jan. 9, 2014

Jan. 9, 2014

Jacek Caban <jacek(a)codeweavers.com> writes: > --- > dlls/devenum/createdevenum.c | 1 + > 1 file changed, 1 insertion(+) It breaks the tests: ../../../tools/runtest -q -P wine -T ../../.. -M dxdiagn.dll -p dxdiagn_test.exe.so provider && touch provider.ok provider.c:84: Test failed: Expected IDxDiagProvider::Initialize to return S_OK, got 80070002 provider.c:88: Test failed: Expected IDxDiagProvider::Initialize to return S_OK, got 80070002 provider.c:125: Test failed: Expected IDxDiagProvider::Initialize to return S_OK, got 80070002 provider.c:128: Tests skipped: IDxDiagProvider::Initialize failed make: *** [provider.ok] Error 3 -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: ws2_32/tests: Add broken for a WSALookupServiceBegin test in NT
by Marvin Jan. 9, 2014

Jan. 9, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4089 Your paranoid android. === wxppro (32 bit sock) === sock.c:557: Test failed: oob_server (174): unexpectedly at the OOB mark: 0 sock: unhandled exception c0000005 at 71AB4047 === w2008s64 (32 bit sock) === sock.c:557: Test failed: oob_server (750): unexpectedly at the OOB mark: 0 sock: unhandled exception c0000005 at 00000000 === w7pro64 (32 bit sock) === sock.c:566: Test failed: oob_server (964): not at the OOB mark: 1 === w2008s64 (64 bit sock) === sock.c:557: Test failed: oob_server (654): unexpectedly at the OOB mark: 0

1 0

Re: [PATCH 5/5] devenum: Added more strict CLSID_VideoCompressorCategory test.
by Marvin Jan. 9, 2014

Jan. 9, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4077 Your paranoid android. === w2008s64 (64 bit devenum) === devenum.c:124: Test failed: mrle codec not found devenum.c:124: Test failed: mrle codec not found === w7pro64 (64 bit devenum) === devenum.c:124: Test failed: mrle codec not found devenum.c:124: Test failed: mrle codec not found

1 0

Re: [PATCH 2/5] ddraw/tests: Add some color key tests for complex surfaces.
by Marvin Jan. 9, 2014

Jan. 9, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4071 Your paranoid android. === w2008s64 (64 bit ddraw7) === ddraw7.c:1621: Test failed: Failed to create surface, hr 0x80004002. === w7pro64 (64 bit ddraw7) === ddraw7.c:1621: Test failed: Failed to create surface, hr 0x80004002.

1 0

Re: [PATCH] oleaut32: Mask out certain features on array copy
by Dmitry Timoshkov Jan. 8, 2014

Jan. 8, 2014

Nikolay Sivov <nsivov(a)codeweavers.com> wrote: > +/* features listed here are not propagated to newly created array or data copy > + created with SafeArrayCopy()/SafeArrayCopyData() */ > +static const USHORT ignored_copy_features = > + FADF_AUTO | > + FADF_STATIC | > + FADF_EMBEDDED | > + FADF_FIXEDSIZE | > + FADF_CREATEVECTOR | > + FADF_DATADELETED; ... > - (*ppsaOut)->fFeatures = psa->fFeatures & ~FADF_CREATEVECTOR; > + (*ppsaOut)->fFeatures = psa->fFeatures & ~(FADF_CREATEVECTOR|ignored_copy_features); ignored_copy_features already includes FADF_CREATEVECTOR. > --- a/dlls/oleaut32/tests/safearray.c > +++ b/dlls/oleaut32/tests/safearray.c > @@ -57,6 +57,14 @@ static BOOL has_i8; > /* Has INT_PTR/UINT_PTR type? */ > static BOOL has_int_ptr; > > +static const USHORT ignored_copy_features[] = > + { > + FADF_AUTO, > + FADF_STATIC, > + FADF_EMBEDDED, > + FADF_FIXEDSIZE > + }; Tested and actually ignored flags do not match. Either the test is not complete or the ignored list includes too much. -- Dmitry.

2 1

Re: [PATCH 2/5] ddraw/tests: Add some color key tests for complex surfaces.
by Stefan Dösinger Jan. 8, 2014

Jan. 8, 2014

Am 08.01.2014 um 10:56 schrieb Henri Verbeet <hverbeet(a)codeweavers.com>: > + for (i = 0; i < 7; ++i) > + { > + hr = IDirectDrawSurface_GetAttachedSurface(mipmap, &caps, &tmp); > + ok(SUCCEEDED(hr), "Failed to get attached surface, i %u, hr %#x.\n", i, hr); > + hr = IDirectDrawSurface_GetColorKey(tmp, DDCKEY_SRCBLT, &color_key); > + ok(hr == DDERR_NOCOLORKEY, "Got unexpected hr %#x, i %u.\n", hr, i); > + IDirectDrawSurface_Release(mipmap); > + mipmap = tmp; > + } What is the 3D rendering and blitting behavior of the mipmap sublevels? Does an application have to set the color key on all surfaces to get proper color keying on a mipmapped texture, or do the sublevels use the color key from the root surface?

2 6

Re: dpnet: Add stubbed interface IDirectPlay8Server
by Michael Stefaniuc Jan. 8, 2014

Jan. 8, 2014

Hello Alistair, On 01/08/2014 06:26 AM, Alistair Leslie-Hughes wrote: > Changelog: > dpnet: Add stubbed interface IDirectPlay8Server > diff --git a/dlls/dpnet/dpnet_private.h b/dlls/dpnet/dpnet_private.h > index 72012cd..5195d51 100644 > --- a/dlls/dpnet/dpnet_private.h > +++ b/dlls/dpnet/dpnet_private.h > @@ -37,6 +37,7 @@ typedef struct IDirectPlay8AddressImpl IDirectPlay8AddressImpl; > typedef struct IDirectPlay8LobbiedApplicationImpl IDirectPlay8LobbiedApplicationImpl; > typedef struct IDirectPlay8PeerImpl IDirectPlay8PeerImpl; > typedef struct IDirectPlay8ThreadPoolImpl IDirectPlay8ThreadPoolImpl; > +typedef struct IDirectPlay8ServerImpl IDirectPlay8ServerImpl; please add the struct directly to server.c as it is used only there. > +static ULONG WINAPI IDirectPlay8ServerImpl_AddRef(IDirectPlay8Server *iface) > +{ > + IDirectPlay8ServerImpl *This = impl_from_IDirectPlay8Server(iface); > + ULONG refCount = InterlockedIncrement(&This->ref); > + > + TRACE("(%p)->(ref before=%u)\n", This, refCount - 1); The standard to which Wine moves is to print the current refcount. See http://wiki.winehq.org/COMGuideline under "COM aggregation" for sample AddRef / Release functions. > +HRESULT DPNET_CreateDirectPlay8Server(LPCLASSFACTORY iface, LPUNKNOWN punkOuter, REFIID riid, LPVOID *ppobj) IClassFactory* and IUnknown* are much nicer. > +{ > + IDirectPlay8ServerImpl *server; > + > + TRACE("(%p, %s, %p)\n", punkOuter, debugstr_guid(riid), ppobj); > + > + server = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(IDirectPlay8ServerImpl)); > + if (!server) > + { > + *ppobj = NULL; > + return E_OUTOFMEMORY; > + } > + server->IDirectPlay8Server_iface.lpVtbl = &DirectPlay8ServerVtbl; > + server->ref = 1; > + > + *ppobj = (LPVOID*)&server->IDirectPlay8Server_iface; This will fail to fail for an unsupported riid. I have added the correct way to the COMGuideline wiki page. thanks bye michael

1 0

Re: dssenh/tests: Added Cylink Mek algorithm definition and crypto provider algorithm tests for dssenh.
by Alexandre Julliard Jan. 8, 2014

Jan. 8, 2014

Marek Chmiel <kcmark(a)gmail.com> writes: > + while(GetLastError() != ERROR_NO_MORE_ITEMS) > + { > + SetLastError(0xdeadbeef); > + dataLen = sizeof(dataBuffer); > + > + result = CryptGetProvParam(hProv, PP_ENUMALGS_EX, dataBuffer, &dataLen, flag); > + ok(result || (!result && GetLastError() == ERROR_NO_MORE_ITEMS), > + "Failed to get parameter, got %x\n", GetLastError()); > + > + pointer = dataBuffer; > + algid = *(ALG_ID *)pointer; pointer += sizeof(ALG_ID); > + defaultLen = *(DWORD *)pointer; pointer += dwIncrement; It doesn't make sense to test last error before the first call, or to test the buffer contents on failure. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: shell32/tests: Add test for SHGetPathFromIDListA (try 3)
by Nikolay Sivov Jan. 8, 2014

Jan. 8, 2014

On 1/8/2014 12:47, Zhenbo Li wrote: > try 3: > Add 'static' in my code > I forgot to call test_SHGetPathFromIDListA() in try2 > Check GetLastError() as well > try 2: > Move code to string.c to avoid creating a new file. > Avoid malloc() and free(). > > I'm working on Bug 33750, so I'll add test cases for > SHGetPathFromIDListA gradually. > --- > dlls/shell32/tests/string.c | 14 ++++++++++++++ > 1 file changed, 14 insertions(+) > +static void test_SHGetPathFromIDListA(void) > +{ > + BOOL ret; > + char pszPath[MAX_PATH]; > + > + ret = SHGetPathFromIDListA(NULL, pszPath); > + ok(!ret, "SHGetPathFromIDListA should fail.\n"); > + ok(GetLastError() == ERROR_PROC_NOT_FOUND, > + "Bad GetLastError = %d\n", GetLastError()); > +} > + You need to set last error to something like 0xdeadbeef, your test suggest that this call doesn't set it. I guess you're trying to test return value only, but it could be interesting to test what happens to path buffer too. P.S. please avoid psz* style prefixes in names.

1 0

Re: shell32/tests: Add test for SHGetPathFromIDListA (try 3)
by Marvin Jan. 8, 2014

Jan. 8, 2014

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://newtestbot.winehq.org/JobDetails.pl?Key=4066 Your paranoid android. === wxppro (32 bit string) === string.c:107: Test failed: Bad GetLastError = 0 === wvista (32 bit string) === string.c:107: Test failed: Bad GetLastError = 0 === w2008s64 (32 bit string) === string.c:107: Test failed: Bad GetLastError = 0 === w7pro64 (32 bit string) === string.c:107: Test failed: Bad GetLastError = 1114 === w2008s64 (64 bit string) === string.c:107: Test failed: Bad GetLastError = 0 === w7pro64 (64 bit string) === string.c:107: Test failed: Bad GetLastError = 1114

1 0

Re: shell32/tests: Add test for SHGetPathFromIDListA
by Dmitry Timoshkov Jan. 8, 2014

Jan. 8, 2014

Zhenbo Li <litimetal(a)gmail.com> wrote: > +void test_SHGetPathFromIDListA(void) > +{ > + BOOL ret; > + char pszPath[MAX_PATH]; > + > + ret = SHGetPathFromIDListA(NULL, pszPath); > + ok(!ret, "SHGetPathFromIDListA should fail.\n"); > +} 1. the test should be 'static'. 2. who is supposed to call this test? 3. it would be ineteresting to see whether the tested API should set an error code or somehow indicate a reason of the failure. -- Dmitry.

2 1