September 2014 - Wine-Devel

Re: d3d10core: Add support for map_type flags D3D10_MAP_READ, D3D10_MAP_WRITE_NO_OVERWRITE and D3D10_MAP_DISCARD.

by Henri Verbeet

On 23 September 2014 12:55, Johannes Brandstätter <jbrandst(a)2ds.eu> wrote: > + switch (map_type) > + { > + case D3D10_MAP_READ: > + wine_map_flags = WINED3D_MAP_READONLY; > + break; > + > + case D3D10_MAP_WRITE_DISCARD: > + wine_map_flags = WINED3D_MAP_DISCARD; > + break; > + > + case D3D10_MAP_WRITE_NO_OVERWRITE: > + wine_map_flags = WINED3D_MAP_NOOVERWRITE; > + break; > + > + default: > + FIXME("Ignoring map_type %#x.\n", map_type); > + } > + } You'll want to turn that into a helper function and use it for the other resource mapping functions as well. See also e.g. wined3d_usage_from_d3d10core() and wined3dformat_from_dxgi_format().

Sept. 24, 2014

Re: [PATCH 4/5] d3d9/tests: Extend focus loss tests.

by Henri Verbeet

On 23 September 2014 23:54, Stefan Dösinger <stefandoesinger(a)gmail.com> wrote: > Am 2014-09-23 13:52, schrieb Henri Verbeet: >> I don't think the tests really support the implementation. For >> example, they don't test multiple swapchains at all, and they >> basically only test if WS_MINIMIZE is set or not as far as window >> styles are concerned. I'd expect at least some basic message and >> style tests along the lines of e.g. test_wndproc() and >> test_window_style(). > Is the attached test what you had in mind? Mode switching is not Something along those lines, yeah. You'll probably want to use a wparam mask instead of multiple expected values, and I'd prefer WA_INACTIVE over 0 in the appropriate places. > I can extend test_window_style as well to check for WS_VISIBLE, but I > don't see a reason to do that given the messages being sent. > You probably mostly care about WS_MINIMIZE, but it's also to verify that there are no other style changes. I imagine this would also be useful if we ever get around to finishing e.g. a WGL extension for switching to/from fullscreen. > Wrt multiple swapchains, since we don't have any tests and support for > multi-head or CreateAdditionalSwapchain in fullscreen mode that'd be a > rather huge detour I'd prefer not to take. > I suspect that adding tests for CreateAdditionalSwapChain() at least would only be a modest amount of work, but without tests, I'd expect that we should only minimize the fullscreen swapchain. Or, put a different way, only the window that we received the message for.

Sept. 24, 2014

Re: [PATCH 1/2] jscript: Add define for JS_E_OUT_OF_MEMORY.

by Qian Hong

On Wed, Sep 24, 2014 at 5:20 PM, Jacek Caban <jacek(a)codeweavers.com> wrote: > Adding private defines that are not used is not really useful. Please at > least add resources. Thanks Jacek, a newer version is sent. -- Regards, Qian Hong - http://www.codeweavers.com

Sept. 24, 2014

Re: [PATCH] mshtml: Add IHTMLAnchorElement::search propertyimplementation(try 3)

by Jacek Caban

Hi Hao, On 09/24/14 11:14, Hao Peng wrote: > +#define test_anchor_search_strange(a,h,n) _test_anchor_search_strange(__LINE__,a,h,n) > +static void _test_anchor_search_strange(unsigned line, IUnknown *unk, const char *search, const char *result) > +{ > + IHTMLAnchorElement *anchor = _get_anchor_iface(line, unk); > + BSTR str; > + HRESULT hres; > + > + if ( ! result) > + result = search; > + > + str = search ? a2bstr(search) : NULL; > + hres = IHTMLAnchorElement_put_search(anchor, str); > + ok_(__FILE__,line)(hres == S_OK, "put_search failed: %08x\n", hres); > + SysFreeString(str); > + > + hres = IHTMLAnchorElement_get_search(anchor, &str); > + ok_(__FILE__,line)(hres == S_OK, "get_search failed: %08x\n", hres); > + if ( ! str) > + /* due to IE5's incorrect behavior, search string without leading "?" is interpreted > + as part of the pathname, and can not be accessed by get_search. */ > + win_skip("incorrect behavior on IE5"); Why do you need this? Those tests are already not ran on IE5 anyway. Jacek

Sept. 24, 2014

Re: [PATCH 1/2] jscript: Add define for JS_E_OUT_OF_MEMORY.

by Jacek Caban

Hi Qian, On 09/24/14 10:45, Qian Hong wrote: > --- a/dlls/jscript/resource.h > +++ b/dlls/jscript/resource.h > @@ -24,6 +24,7 @@ > > #define IDS_TO_PRIMITIVE 0x0001 > #define IDS_INVALID_CALL_ARG 0x0005 > +#define IDS_OUT_OF_MEMORY 0x0007 Adding private defines that are not used is not really useful. Please at least add resources. Jacek

Sept. 24, 2014

Re: libwine: Respect TMPDIR environment variable

by Alexandre Julliard

André Hentschel <nerv(a)dawncrow.de> writes: > @@ -36,7 +36,7 @@ > #include "wine/library.h" > > static const char server_config_dir[] = "/.wine"; /* config dir relative to $HOME */ > -static const char server_root_prefix[] = "/tmp/.wine"; /* prefix for server root dir */ > +static const char server_root_prefix[] = "/.wine"; /* prefix for server root dir */ > static const char server_dir_prefix[] = "/server-"; /* prefix for server dir */ > > static char *bindir; > @@ -220,15 +220,19 @@ static char *get_runtime_argvdir( const char *argv0 ) > /* initialize the server directory value */ > static void init_server_dir( dev_t dev, ino_t ino ) > { > + const char* tmpdir; > char *p, *root; > > + if (!(tmpdir = getenv("TMPDIR"))) tmpdir = "/tmp"; The server dir needs to be unique for a given prefix, so it cannot depend on environment variables. -- Alexandre Julliard julliard(a)winehq.org

Sept. 24, 2014

Patches for adding IRichEditOle/ITextDocument support for ITextServices::QueryInterface ask for review

by Jactry Zeng

Hi folks, I wrote two patches for adding IRichEditOle/ITextDocument support for ITextServices::QueryInterface(for bug 17042[1]), but I'm no sure is it the correct way. I need your review and comment! I will really appreciate it! There are some tests for ITextServices::QueryInterface in dlls/riched20/txtsrv.c: test_QueryInterfaces(): 1. refcount = get_refcount((IUnknown *)txtserv); ok(refcount == 1, "got wrong ref count: %d\n", refcount); /* IID_IRichEditOle */ hres = ITextServices_QueryInterface(txtserv, &IID_IRichEditOle, (void **)&txtsrv_reOle); ok(hres == S_OK, "ITextServices_QueryInterface\n"); refcount = get_refcount((IUnknown *)txtserv); ok(refcount == 2, "got wrong ref count: %d\n", refcount); >From this case we know ref count of ITextServices will increase after get a IRichEditOle interface from ITextServices::QueryInterface. ITextServices::QueryInterface(&IID_ITextDocument) is also similar. 2. refcount = get_refcount((IUnknown *)txtserv); ok(refcount == 2, "got wrong ref count: %d\n", refcount); if (SUCCEEDED(hres)) { refcount = get_refcount((IUnknown *)txtsrv_reOle); ok(refcount == 2, "got wrong ref count: %d\n", refcount); hres = IRichEditOle_QueryInterface(txtsrv_reOle, &IID_ITextDocument, (void **)&txtDoc); ok(hres == S_OK, "IRichEditOle_QueryInterface: 0x%08x\n", hres); refcount = get_refcount((IUnknown *)txtserv); ok(refcount == 3, "got wrong ref count: %d\n", refcount); refcount = get_refcount((IUnknown *)txtsrv_reOle); ok(refcount == 3, "got wrong ref count: %d\n", refcount); >From this case we know when a IRichEditOle was got by ITextServices::QueryInterface(&IID_IRichEditOle) then we used IRichEditOle::QueryInterface to get another interface, the ref count of ITextServices will increase too. For implementing this feature I added IUnknown *parent in IRichEditOleImpl and passed ITextServices interface into CreateIRichEditOle when we call it in ITextServices::QueryInterface: dlls/riched20/richole.c ---snip--- typedef struct IRichEditOleImpl { IRichEditOle IRichEditOle_iface; ITextDocument ITextDocument_iface; LONG ref; ME_TextEditor *editor; ITextSelectionImpl *txtSel; IOleClientSiteImpl *clientSite; struct list rangelist; IUnknown *parent; } IRichEditOleImpl; ---snip--- dlls/riched20/txtsrv.c ---snip--- else if (IsEqualIID(riid, &IID_IRichEditOle) || IsEqualIID(riid, &IID_ITextDocument)) { if (!This->editor->reOle) CreateIRichEditOle(This->editor, (LPVOID *)&(This->editor->reOle), iface); IRichEditOle_QueryInterface(This->editor->reOle, riid, (void **)ppv); return S_OK; } else { ---snip--- 3. By some other tests, I found when a ITextServices was got by IRichEditOle::QueryInterface, then we use ITextServices::QueryInterface to got another interface, the ref count of IRichEditOle will increase too. Thanks! [1] https://bugs.winehq.org/show_bug.cgi?id=17042 -- Regards, Jactry Zeng

Sept. 24, 2014

Re: [PATCH 1/2] riched20: Stub for ITextFont interface and implement ITextRange::GetFont. (try 3)

by Jactry Zeng

Hi Nikolay, Thanks for your review! I will have another try. 2014-09-23 23:01 GMT+08:00 Nikolay Sivov <bunglehead(a)gmail.com>: > +typedef struct ITextFontImpl { >> + ITextFont ITextFont_iface; >> + LONG ref; >> + >> + ITextRangeImpl *txtRge; >> + ITextSelectionImpl *txtSel; >> +} ITextFontImpl; >> + >> > I think 'range' and 'selection' would be more readable, but what's more > important you don't use 'txtSel' at all. > > +static HRESULT WINAPI ITextFont_fnQueryInterface(ITextFont *me, REFIID >> riid, void **ppvObj) >> > If we want some unification, it's usually called 'iface', not 'me'. > > + *ppvObj = me; >> > That's not how it should look. > > +static IRichEditOleImpl *get_reOle(ITextFontImpl *txtFont) >> +{ >> + if (txtFont->txtRge) >> + return txtFont->txtRge->reOle; >> + else >> + return txtFont->txtSel->reOle; >> +} >> > Please use a better naming, like 'get_reole_from_font()' or something like > that. > > + FIXME("not implemented: %p\n", This); >> + return E_NOTIMPL; >> > This should give a proper trace line with arguments, and it should appear > on every call. > > + ITextFontImpl *This = impl_from_ITextFont(me); >> + if (pValue) >> + return E_INVALIDARG; >> + if (!get_reOle(This)) >> + return CO_E_RELEASED; >> + >> + FIXME("Stub\n"); >> + *pValue = tomFalse; >> + return S_OK; >> > That's another example of tracing done wrong. > > static HRESULT WINAPI ITextRange_fnGetFont(ITextRange *me, ITextFont >> **pFont) >> { >> ITextRangeImpl *This = impl_from_ITextRange(me); >> + ITextFontImpl *txtFont = NULL; >> + HRESULT hres; >> + >> if (!This->reOle) >> return CO_E_RELEASED; >> - FIXME("not implemented %p\n", This); >> - return E_NOTIMPL; >> + TRACE("%p\n", This); >> > And another. > > > -- Regards, Jactry Zeng

Sept. 24, 2014

Re: [PATCH 4/5] d3d9/tests: Extend focus loss tests.

by Stefan Dösinger

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 2014-09-23 13:52, schrieb Henri Verbeet: > I don't think the tests really support the implementation. For > example, they don't test multiple swapchains at all, and they > basically only test if WS_MINIMIZE is set or not as far as window > styles are concerned. I'd expect at least some basic message and > style tests along the lines of e.g. test_wndproc() and > test_window_style(). Is the attached test what you had in mind? Mode switching is not included yet. Windows 7 and Windows Vista don't agree on the WM_WINDOWPOSCHANGING / WM_WINDOWPOSCHANGED parameters, so I'm not checking them. I can extend test_window_style as well to check for WS_VISIBLE, but I don't see a reason to do that given the messages being sent. Wrt multiple swapchains, since we don't have any tests and support for multi-head or CreateAdditionalSwapchain in fullscreen mode that'd be a rather huge detour I'd prefer not to take. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJUIewwAAoJEN0/YqbEcdMwsZIP/2hTwfUuPLSKSoC8MQ4dtEju /awDAyXOtKLwYl1FzyTYr8yugqXpPiJk0PEuId2HppVHVnbhVhH16eOt+iCPDpOy J+9rzt08OaGEPp5tVpqExI7aqbMdb+sxNApqi1rRwzmLyl6fmSFJuNj8LRmu2hTR EuIrTSqOVZwXKVXOZJ7OdeYG6Vm7TefLT8AAIqeI6FmPcnwBg350nXa13V01OYlO tzsGzQSK3IPA/urL6BVZea6Yuh4++mIq5/CFsTwfxXqjbnFKy+nUvtt4F7icJpe2 YgHb02C6eVm3vvwyatO/gfxqAjMhF3AoUo2bVnbLA55bM45H7EjDlBm6xAZuIevE 4qCl/Rdk9dMNTjI8e/s5PjGkJpJgn+LvWFFJBG6AmE5zJqGKswRhMjwLPP8WdVEl mUNtG3QutggqWfK3h2IXtA27GhUR8z8yi8oxLt4l/Oh2X18iTD1BjXNkFPJrKnIL Tr7eafhMW5R0GwNDfA9ImKMS4IIkWylhrIcIKhuDKpGUz3j5QqQ+TkHOwPob6xez kx+5B/9fRqVcepbEpfw7UHxcbW4P3l4Q/kkP02cRSETAKetO8vn17jA8GMzGfdGV mURE0QLciZ/d4/4kkjwv/fvLjilaW+qWi5FVgSWhOFg3ay2jrLadmCFQfnOmgmfG 0QEZ+3dkS2TAqb+MyCN+ =jmOV -----END PGP SIGNATURE-----

Sept. 23, 2014

Re: Question about implementing a test for BCryptGetFipsAlgorithmMode

by Kai Blaschke

> Kai, out of curiosity, is there an app that depends on being able to > enable FIPS mode? > --Juan The only real-life implementation I know about using the flag and actually implementing FIPS is the MS .NET framework. Any application making use of functionality in the System.Security.Cryptography namespace will be limited to algorithms allowed in FIPS if the setting is enabled. You can try this by enabling FIPS on Windows and playing Terraria, for example. In this case, the application depends on FIPS mode being disabled. Kai

Sept. 23, 2014