March 2008 - Wine-devel - list.winehq.org

Question about function "HTMLDocument_write"
by Ivan Sinitsin 13 Mar '08

13 Mar '08

On Monday the patch realizing this function has been accepted. It works perfectly, but hyperlinks do not work. If I makes this function that way: static HRESULT WINAPI HTMLDocument_write(IHTMLDocument2 *iface, SAFEARRAY *psarray) { HRESULT hres; VARIANT *pvar; IHTMLElement *pbody; BSTR vtext; int i; vtext = SysAllocString(L""); HTMLDocument_get_body(iface,&pbody); hres = IHTMLElement_get_innerHTML(pbody, &vtext); if (FAILED(hres)) { vtext = SysAllocString(L""); } hres=SafeArrayAccessData(psarray, (void **)&pvar); if (FAILED(hres)) return hres; for (i=0; i < psarray->rgsabound[0].cElements; i++) { hres = VarBstrCat(vtext, V_BSTR(&(pvar[i])), &vtext); if (FAILED(hres)) { break; } } IHTMLElement_put_innerHTML(pbody,vtext); SysFreeString(vtext); hres = SafeArrayUnaccessData(psarray); return S_OK; } hyperlinks works good. Why so occurs?

2 2

Re: msi: Ignore the custom action type 51 if the source field is empty
by Paul Vriens 13 Mar '08

13 Mar '08

James Hawkins wrote: > Hi, > > Fixes bug 11891. http://bugs.winehq.org/show_bug.cgi?id=11891 > > Changelog: > * Ignore the custom action type 51 if the source field is empty. > > dlls/msi/custom.c | 3 ++ > dlls/msi/tests/install.c | 65 ++++++++++++++++++++++++++++++++++++++++++++++ > 2 files changed, 68 insertions(+), 0 deletions(-) > > > > ------------------------------------------------------------------------ > > Hi James, I was looking at the sudden huge amount of failures for the msi/package tests on Windows: http://test.winehq.org/data/200803121000/ The issue turned out to be with the above patch for the install tests. If I run the install tests a 'MSITEST' package will stay around (can been seen in 'Add/Remove Programs"). If I remove MSITEST the package tests will succeed again. (Reverting the above patch makes the package tests pass again as well, when run after the install tests of course). The thing that surprises me is that we don't see the same failures when running the package tests on Wine? Is this because we have 39 TODO's in there? -- Cheers, Paul.

1 0

re: Running services in WINE?
by Dan Kegel 12 Mar '08

12 Mar '08

Yes, services are implemented now. If you run into a problem, please file a bug.

1 0

Vista and SHFileOperation tests for shell32
by Paul Vriens 12 Mar '08

12 Mar '08

Hi, I was trying to fix the issues with the user32/listbox tests when I found that we have a lot of leftover files from the shell32/shlfileop test. This happens because I get about a gazillion dialogs asking me what to do during the shlfileop tests and I usually tend to click Cancel or Skip. The first test that fails for me is where we delete the directory before the file in that directory. This silently succeeds on everything before Vista but fails on Vista and W2K8. Adding any of the silencing flags doesn't make a difference on Vista but make it fail on all the other Windows versions. What road should I take for this: a) Add some delete_all_possible_leftover_files_and_directories function at the end of shlfileop.c ? b) Start making a distinction between Vista/W2K8 and pre-Vista ? c) ???? The first one obviously doesn't fix the test failures. -- Cheers, Paul.

1 0

Re: [1/4] qmgr: Add critical sections for Jobs and Files.
by Alexandre Julliard 12 Mar '08

12 Mar '08

Dan Hipschman <dsh(a)linux.ucla.edu> writes: > @@ -25,6 +25,7 @@ WINE_DEFAULT_DEBUG_CHANNEL(qmgr); > > static void BackgroundCopyFileDestructor(BackgroundCopyFileImpl *This) > { > + This->owner->lpVtbl->Release((IBackgroundCopyJob *) This->owner); Please use the appropriate COM macros instead of accessing lpVtbl directly. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: Free apps bundled with msvcp71?
by Remco 12 Mar '08

12 Mar '08

----- Original Message ---- > From: Dan Kegel <dank(a)kegel.com> > Subject: Re: Free apps bundled with msvcp71? > > I'm still looking for a legal download of some app > that just happens to come with msvcp71.dll. > Ideally it'll be some small app with a trivial installer > that can be run noninteractively, suitable for use inside winetricks. Maybe winetricks could become a w32 application. Make it an applicationthat "installs commonly omitted dlls and frameworks on Windows". That way,you could use some of the dlls directly: those dlls that only allowfree redistribution when bundled with Windows-oriented applications. Nothing in the application, especially the name,should refer to Wine of course; that alone would make it orientedto non-Windows platforms. At least until there is a plausible use for Wine on Windows. Remco __________________________________________________________ Sent from Yahoo! Mail. The World's Favourite Email http://uk.docs.yahoo.com/nowyoucan.html

1 0

Can we have a server component in bugzilla?
by Dan Kegel 12 Mar '08

12 Mar '08

There have been several server-related bugs lately, yet we have no server component. Seems like we need one.

2 4

Re: [2/2] wined3d: Don't translate by a full half pixel
by Stefan Dösinger 12 Mar '08

12 Mar '08

Am Mittwoch, 12. März 2008 09:46:25 schrieb Chris Robinson: I am not yet conviced that those patches are a good idea. Some time ago I found that I need the full half pixel to make DCT happy. Furthermore I see different behavior among drivers, e.g. my r200 driver doesn't need any correction(it rather hurts instead), and the nvidia driver behaves odd on very small render targets. As I said on IRC, I am afraid that we have to do those position fixes per driver. In the extension detection we have to draw a few points onto the back buffer and read back the results, and then configure the offsets accordingly. As I keep saying, we need proper tests for that. Your test to test the corner points is a good start, but we also have to test offscreen targets of different sizes(e.g. 1x1, 4x4, 16x16, ...), and we have to test all 3 codepaths(rhw, fixed function, shader).

1 0

Re: Hans Leidekker : wininet: Make sure not to overwrite any caller supplied authorization header.
by Robert Shearman 12 Mar '08

12 Mar '08

Alexandre Julliard wrote: > Module: wine > Branch: master > Commit: b069ef4268e7056856ce6714714d929165f24fc8 > URL: http://source.winehq.org/git/wine.git/?a=commit;h=b069ef4268e7056856ce67147… > > Author: Hans Leidekker <hans(a)it.vu.nl> > Date: Fri Feb 1 14:40:15 2008 +0100 > > wininet: Make sure not to overwrite any caller supplied authorization header. > > --- > > dlls/wininet/http.c | 36 ++++++++++-------------------------- > dlls/wininet/tests/http.c | 5 ++++- > 2 files changed, 14 insertions(+), 27 deletions(-) > > diff --git a/dlls/wininet/http.c b/dlls/wininet/http.c > index 65b86fb..a339c86 100644 > --- a/dlls/wininet/http.c > +++ b/dlls/wininet/http.c > @@ -1187,9 +1187,11 @@ static UINT HTTP_DecodeBase64( LPCWSTR base64, LPSTR bin ) > * > * Insert or delete the authorization field in the request header. > */ > -static BOOL HTTP_InsertAuthorizationForHeader( LPWININETHTTPREQW lpwhr, struct HttpAuthInfo *pAuthInfo, LPCWSTR header ) > +static BOOL HTTP_InsertAuthorization( LPWININETHTTPREQW lpwhr, LPCWSTR header, BOOL first ) > { > WCHAR *authorization = NULL; > + struct HttpAuthInfo *pAuthInfo = lpwhr->pAuthInfo; > + DWORD flags; > > if (pAuthInfo && pAuthInfo->auth_data_len) > { > @@ -1222,35 +1224,17 @@ static BOOL HTTP_InsertAuthorizationForHeader( LPWININETHTTPREQW lpwhr, struct H > > TRACE("Inserting authorization: %s\n", debugstr_w(authorization)); > > - HTTP_ProcessHeader(lpwhr, header, authorization, > - HTTP_ADDHDR_FLAG_REPLACE | HTTP_ADDHDR_FLAG_REQ); > + /* make sure not to overwrite any caller supplied authorization header */ > + flags = HTTP_ADDHDR_FLAG_REQ; > + flags |= first ? HTTP_ADDHDR_FLAG_ADD_IF_NEW : HTTP_ADDHDR_FLAG_REPLACE; > > - HeapFree(GetProcessHeap(), 0, authorization); > + HTTP_ProcessHeader(lpwhr, header, authorization, flags); > > + HeapFree(GetProcessHeap(), 0, authorization); > return TRUE; > } > > /*********************************************************************** > - * HTTP_InsertAuthorization > - * > - * Insert the authorization field in the request header > - */ > -static BOOL HTTP_InsertAuthorization( LPWININETHTTPREQW lpwhr ) > -{ > - return HTTP_InsertAuthorizationForHeader(lpwhr, lpwhr->pAuthInfo, szAuthorization); > -} > - > -/*********************************************************************** > - * HTTP_InsertProxyAuthorization > - * > - * Insert the proxy authorization field in the request header > - */ > -static BOOL HTTP_InsertProxyAuthorization( LPWININETHTTPREQW lpwhr ) > -{ > - return HTTP_InsertAuthorizationForHeader(lpwhr, lpwhr->pProxyAuthInfo, szProxy_Authorization); > -} > - > -/*********************************************************************** > * HTTP_DealWithProxy > */ > static BOOL HTTP_DealWithProxy( LPWININETAPPINFOW hIC, > @@ -2621,8 +2605,8 @@ BOOL WINAPI HTTP_HttpSendRequestW(LPWININETHTTPREQW lpwhr, LPCWSTR lpszHeaders, > lpwhr->hdr.dwFlags & INTERNET_FLAG_KEEP_CONNECTION ? szKeepAlive : szClose, > HTTP_ADDHDR_FLAG_REQ | HTTP_ADDHDR_FLAG_REPLACE); > > - HTTP_InsertAuthorization(lpwhr); > - HTTP_InsertProxyAuthorization(lpwhr); > + HTTP_InsertAuthorization(lpwhr, szAuthorization, !loop_next); > + HTTP_InsertAuthorization(lpwhr, szProxy_Authorization, !loop_next); > > /* add the headers the caller supplied */ > if( lpszHeaders && dwHeaderLength ) > diff --git a/dlls/wininet/tests/http.c b/dlls/wininet/tests/http.c > index 8bb7cca..88369cf 100644 > --- a/dlls/wininet/tests/http.c > +++ b/dlls/wininet/tests/http.c > @@ -1502,7 +1502,10 @@ static void test_header_handling_order(int port) > request = HttpOpenRequest(connect, NULL, "/test3", NULL, NULL, types, INTERNET_FLAG_KEEP_CONNECTION, 0); > ok(request != NULL, "HttpOpenRequest failed\n"); > > - ret = HttpSendRequest(request, authorization, ~0UL, NULL, 0); > + ret = HttpAddRequestHeaders(request, authorization, ~0UL, HTTP_ADDREQ_FLAG_ADD); > + ok(ret, "HttpAddRequestHeaders failed\n"); > + > + ret = HttpSendRequest(request, NULL, 0, NULL, 0); > ok(ret, "HttpSendRequest failed\n"); > > status = 0; > This patch simultaneously breaks proxy authentication and NTLM authentication. Proxy authentication is broken because the helper function is removed that passes lpwhr->pProxyAuthInfo into HTTP_InsertAuthentication. NTLM authentication is broken because first is false when we try to send the authenticate message (i.e. the 3rd leg) and so the data for the negotiate message (i.e. the first leg) is sent instead as the field is not overridden. Is this trying to fix a real app? If not, why isn't the code that is using the behaviour using INTERNET_FLAG_NO_AUTH? -- Rob Shearman

1 0

Re: [PATCH 1/4] avifil32: Fix copy/paste error.
by Stefan Dösinger 12 Mar '08

12 Mar '08

I think the patches suffer from mail client line wrapping, I cannot apply them: patch -p1 < /dev/shm/kenpatch/\[PATCH\ 1_4\]\ avifil32\:\ Fix\ copy_paste\ error. patching file dlls/avifil32/acmstream.c patch: **** malformed patch at line 67: LONG start,

2 1