June 2006 - Wine-devel - list.winehq.org

Re: [MSI 1/4] add streams table
by Mike McCormack 21 Jun '06

21 Jun '06

Hi, Thanks for submitting the patch, and keeping the coding style consistent! Andrey Turkin wrote: > This patch adds virtual _Streams table to MSI because native MSI > maintains such table > > ChangeLog: > virtual _Streams table added > +static UINT STREAMS_fetch_stream( struct tagMSIVIEW *view, UINT row, UINT col, IStream **stm ) > +{ .... > + > + /* > + * The column marked with the type stream data seems to have a single number > + * which references the column containing the name of the stream data > + * > + * Fetch the column to reference first. > + */ > + r = view->ops->fetch_int( view, row, col, &ival ); I'm not so keen on the way this is done. You seem to have duplicated alot of code from TABLE_fetch_stream here. > + if( r != ERROR_SUCCESS ) > + { > + ERR("1: %d\n", r); > + return r; > + } > + > + ERR("fetched %d\n", ival); And forgotten to remove your debug code. > +MSIVIEWOPS streams_ops = > +{ > + TABLE_fetch_int, > + STREAMS_fetch_stream, > + TABLE_set_int, > + TABLE_insert_row, > + STREAMS_execute, > + TABLE_close, > + TABLE_get_dimensions, > + TABLE_get_column_info, > + TABLE_modify, > + TABLE_delete, > + TABLE_find_matching_rows > +}; The streams table is just another table. How about generating an MSITABLE structure containing the data rather than treat it as a special case in here? Mike

2 1

Re: [WiKI] Common site menu 4/4
by Tony Lambregts 21 Jun '06

21 Jun '06

Is there anything wrong with this patch? Change log: update menus for all sites to have a common "WineHQ Menu". files changed: winehq/winehq.py ? winehq/winehq ? winehq/img/logo.png Index: winehq/winehq.py =================================================================== RCS file: /cvsroot/wine/wiki/winehq/winehq.py,v retrieving revision 1.3 diff -u -r1.3 winehq.py --- winehq/winehq.py 16 May 2005 03:57:10 -0000 1.3 +++ winehq/winehq.py 3 Jun 2006 20:09:15 -0000 @@ -271,6 +271,22 @@ ''' return html % args + def winepanel(self, d): + """ Create wiki panel """ + _ = self.request.getText + html = [ + u'<div class="sidepanel">', + self.sidetitle(_("WineHQ Menu")), + u' <div class="sidecontent">', + u' <li><a href="http://www.winehq.org/"> WineHQ</a></li>', + u' <li><a href="http://appdb.winehq.org/"> AppDB</a></li>', + u' <li><a href="http://bugs.winehq.org/"> Bugzilla</a></li>', + u' <li><a href="http://wiki.winehq.org/"> Wine Wiki</a></li>', + u' </div>', + u'</div>', + ] + return u'\n'.join(html) + def wikipanel(self, d): """ Create wiki panel """ _ = self.request.getText @@ -351,6 +367,7 @@ # Sidebar u'<div id="sidebar">', + self.winepanel(d), self.wikipanel(d), self.pagepanel(d), self.userpanel(d),

3 2

Wine talk in LA june 20th,22nd. Slide check, anyone?
by Dan Kegel 20 Jun '06

20 Jun '06

I'm giving a short talk on Wine at lula.org tonight and again at aitp-la.org Thursday night. The audience at aitp-la is IT professionals who don't really know much about Wine or Linux, so I'm trying for a gentle, practical talk. I'll demo a few apps (Office '97, Picasa, Firefox, Kid Pix, and Visual C++) and I'll even run an app that doesn't work too well yet (Visual Basic 6's IDE). The slides, such as they are, are up at kegel.com/wine/wine-2006-june-talk.ppt kegel.com/wine/wine-2006-june-talk.pdf (I usually use magicpoint, but for this talk, I used Office 97's powerpoint on the theory that one should eat one's own dogfood...) Corrections/improvements welcome. - Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

1 0

riched slowdown
by James Hawkins 20 Jun '06

20 Jun '06

Hi, I just did a +relay log of the ie6setup.exe installation. I had to ctrl-c after the log grew to 3.5gb and the initial screen hadn't been displayed yet. tail -n 1000 relay.log showed call after call to WideCharToMultibyte before I finally killed the install. Here is a sample of one of the calls: WideCharToMultiByte(000004e4,00000000,40895ae4 L"S\5524",00000001,40895ae7,00000001,00000000,40895ae8) So the length of the "string" we're converting is 1. Not exactly efficient. grep'ing the log for WideCharToMultiByte reveals 39426680 calls, and it wasn't even done loading the EULA. I'm pretty sure this is a recent bug in riched, as overriding riched20 and riched32 gets rid of the problem. -- James Hawkins

1 0

Re: lz32: remove dead code from the LZOpenFileW test.
by Detlef Riekenberg 20 Jun '06

20 Jun '06

Saulius Krasuckas wrote: > + ok(file >= 0, "LZOpenFileW failed on switching to a compressed file name\n"); .. > + ok(test.nErrCode == 0, "LZOpenFileW set test.nErrCode to %d\n", > + test.nErrCode); Did you mean "test.nErrCode == ERROR_SUCCESS"? You test the Errorcode here, but you did not use SetLastError() before LZOpenFileW. > + ok(lstrcmpA(test.szPathName, expected) == 0, > + "LZOpenFileW returned '%s', but was expected to return '%s'\n", > + test.szPathName, expected); This is strange: An ANSI-Compare for an UNICODE-Function. After reading the source and OFSTRUCT, this is correct. I suggest a command before the above "ok()". An extra Patch with the WINAPI-Documentation for the Implementation would be nice. This should include a note, that the ANSI-String "szPathName" in "OFSTRUCT" is the correct Parameter for the UNICODE-Implementation. ---- cut ---- This part from your previous Patch is already in the tree, but: > > + /* Delete the file then make sure it doesn't exist anymore. */ > + file = LZOpenFileW(filename_W, &test, OF_DELETE); > + ok(file >= 0, "LZOpenFileA failed on delete\n"); > + retval = GetFileAttributesW(filename_W); > + ok(retval == INVALID_FILE_ATTRIBUTES, > + "GetFileAttributesA succeeded on deleted file\n"); The Messages in the "W"-Tests have ANSI Function-Names -- By By ... ... Detlef

2 1

appdb: makeSafe improvement
by Vitaly Lipatov 20 Jun '06

20 Jun '06

After last security fixes in AppDB there is introduced makeSafe function with one parameter. Typical using of makeSafe: $aClean['sub'] = makeSafe($_REQUEST['sub']); When $_REQUEST does not contain 'sub', we get "Notice: Undefined index:" if mod_php in strict mode (do not permit undefined variables). What about use makeSafe in the form $aClean['sub'] = makeSafe($_REQUEST,'sub'); Then we can use isset inside makeSafe for additional checking. -- Vitaly Lipatov, ALT Linux Team Russia, Saint-Petersburg, www.etersoft.ru

1 0

Winelib Getting Started 1.3.2. Test Drive
by Dee Ayy 20 Jun '06

20 Jun '06

Hello, Would a wine developer please revisit this page http://www.winehq.org/site/docs/winelib-guide/winelib-getting-started and modify it to have a step by step guide to take a Linux box without wine nor winelib to a running notepad exe built by the newbie (and not downloaded)? Or if you are a wine-user, please provide such steps to this list. As a newbie, the statement "It can be found in the programs subdirectory." had me lost. I don't know if the steps I took are correct, but somehow I found http://www.winehq.org/site/cvs. Perhaps the winelib-getting-started page could be modified to direct a newbie how to download the "test drive" at this point. Then I only checked out wine/programs/notepad since the statement "(but read the next section before actually doing this!)" intimidated me on the cvs page. The first error after a make was "dialog.c:27:21: error windows.h: No such file or directory". Then I had the brilliant idea to checkout all of wine, even though I was intimidated initially by the statement "(but read the next section before actually doing this!)". The make still failed with the same error, even though I now have the wine/include directory. Not knowing any better, I modified the Makefile from INCLUDE_PATH = -I. to INCLUDE_PATH = -I.:/my/path/to/wine/include. I retried make and get the same error. I'm stumped. The 1.3.3 Step by step guide was too general for me. But maybe that's because I needed section 1.3.2 to be more detailed for the task at hand, building notepad. And perhaps 1.3.3 can be left alone for more general information. Please advise. Thanks, D

4 4

Re: msi: Fix handling of the no-op identifier in the Directory table [resend]
by Alexandre Julliard 20 Jun '06

20 Jun '06

"James Hawkins" <truiken(a)gmail.com> writes: > Is there anything wrong with this patch? > > Changelog: > * Fix handling of the no-op identifier in the Directory table. You are assigning a const string to a non-const variable, that causes warnings. -- Alexandre Julliard julliard(a)winehq.org

2 2

Re: [WINED3D 3/3] Fix fog in the case of vertex shaders
by Ivan Gyurdiev 20 Jun '06

20 Jun '06

On 6/19/06, Jason Green <jave27(a)gmail.com> wrote: > > - Fixes both ARB and GLSL shaders that use fog. > Is it really necessary to keep track of whether the shader uses fog or not? What happens if you disable this post-processing for a shader that doesn't write to the fog register? Also, how come the ARB version forces the fog value to be > 0, but the GLSL one has both lower and upper limit (clamp to 0,1) ? Shouldn't you use saturate (_SAT) to achieve the same effect?

2 1

Re: Win64 patch 1/13
by Mike McCormack 20 Jun '06

20 Jun '06

Ge van Geldorp wrote: > +SPEC_SRC32 = $(BASEMODULE).spec > +SPEC_SRC64 ?= $(SPEC_SRC32) Not sure we want seperate spec files. In any case, ?= doesn't look portable. > -ULONG WINAPI MAPILogon(ULONG uiparam, LPSTR profile, LPSTR password, > +ULONG WINAPI MAPILogon(ULONG_PTR uiparam, LPSTR profile, LPSTR password, > FLAGS flags, ULONG reserved, LPLHANDLE session) > --- programs/wordpad/wordpad.c 14 Feb 2006 16:48:40 -0000 1.1 > +++ programs/wordpad/wordpad.c 19 Jun 2006 17:34:48 -0000 I thought I already fixed these... I don't think these will apply to the current git tree. > --- dlls/ntdll/tests/generated.c 3 Jul 2005 11:23:30 -0000 1.23 > +++ dlls/ntdll/tests/generated.c 19 Jun 2006 17:20:17 -0000 You should be changing tools/winapi/winapi_test*, not the generated files. The heap changes look useful, I will check them out when I get back from my holiday. Mike

2 4