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

Re: Dinput8: DirectInput8Create rewrite
by Roderick Colenbrander 15 May '06

15 May '06

> On 5/15/06, Roderick Colenbrander <thunderbird2k(a)gmx.net> wrote: > > > > Are you sure it is this patch that causes the regression and not > Vitaly's > > latest one? This patch changes the way dinput8 objects are created and > it > > does this similar to windows. Could you figure out what part breaks it? > > > > Yes, I have tried without Vitaly's. CoCreateInstance is failing: > > err:ole:CoCreateInstance apartment not initialised > err:dinput:DirectInput8Create CoCreateInstance failed with hr = > -2147221008 > err:ole:CoGetClassObject class {ef985e71-d5c7-42d4-ba4d-2d073e2e96f4} > not registered > err:ole:CoGetClassObject class {ef985e71-d5c7-42d4-ba4d-2d073e2e96f4} > not registered > err:ole:CoGetClassObject no class object > {ef985e71-d5c7-42d4-ba4d-2d073e2e96f4} could be created for context > 0x3 > > Is this because I am missing registry entries? > > Jesse > Yes, that's the reason. I forgot that I still need to send the patch for the registry settings. Basicly what you need to add is this: REGEDIT4 [HKEY_CLASSES_ROOT\CLSID\{25E609E4-B259-11CF-BFC7-444553540000}] @="Microsoft DirectInput8" [HKEY_CLASSES_ROOT\CLSID\{25E609E4-B259-11CF-BFC7-444553540000}\InProcServer32] @="C:\\windows\\system32\\dinput8.dll" "ThreadingModel"="Both" -- Echte DSL-Flatrate dauerhaft f�r 0,- Euro*! "Feel free" mit GMX DSL! http://www.gmx.net/de/go/dsl

1 0

Re: riched20: Implement EM_SETUNDOLIMIT and its conformance tests, take 4
by Dimi Paun 15 May '06

15 May '06

On Mon, May 15, 2006 1:22 pm, Matt Finnicum wrote: > Hopefully, i've managed to do this one right. It is, but why not use STACK_SIZE_DEFAULT here as well: + ed->nUndoLimit = 100; /* 100 is the proper default */ -- Dimi Paun <dimi(a)lattica.com> Lattica, Inc.

1 0

Re: Starred Chats Sent Mail Drafts (2) All Mail Spam (2510) Trash Contacts Labels battlezone class 05-06 facebook mfinnic2@uiuc.edu myself Summer of Code wine-appdb (16) wine-bugs wine-devel (10) wine-patches (384) wtf-trafton Edit labels Invite a friend
by Dimi Paun 15 May '06

15 May '06

On Mon, May 15, 2006 11:28 am, Matt Finnicum wrote: > Hello! > > Patch updated w/ Dimi Paun's advice (using #define's for the default > and max values), and slightly cleaned up. Well, not quite (maybe you've sent the wrong patch): + case EM_SETUNDOLIMIT: + { + if ((int)wParam < 0) + editor->nUndoLimit = 100; + else if ((int)wParam > 10000) + /* This should be high enough. it keeps wine from getting killed + from hogging memory. Windows allocates all this memory at once, so + no program would realisticly set a value this high */ + editor->nUndoLimit = 10000; + else + editor->nUndoLimit = wParam; + -- Dimi Paun <dimi(a)lattica.com> Lattica, Inc.

1 0

Re: Starred Chats ... blah blah blah
by Matt Finnicum 15 May '06

15 May '06

This was *supposed* to have been titled: "riched20: Implement EM_SETUNDOLIMIT and its conformance tests, take 3" Sorry! --Matt On 5/15/06, Matt Finnicum <mattfinn(a)gmail.com> wrote: > Hello! > > Patch updated w/ Dimi Paun's advice (using #define's for the default > and max values), and slightly cleaned up. > > Thanks, > --Matthew Finnicum > > Changelog: > riched20: Implemented conformance tests for EM_SETUNDOLIMIT() > riched20: Implemented EM_SETUNDOLIMIT() > > editor.c | 21 +++++++++++++++--- > editstr.h | 4 ++- > tests/editor.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > undo.c | 27 ++++++++++++++++++++++- > 4 files changed, 112 insertions(+), 5 deletions(-) > > >

1 0

Re: dnsapi: Switch from libresolv to libbind for DNS support.
by Marcus Meissner 15 May '06

15 May '06

On Sat, May 13, 2006 at 05:10:55PM +0200, Hans Leidekker wrote: > > This patch presents an alternative to the one I sent a while ago > that adds two source files from libbind. The advantage of this > alternative is that it brings DNS message parsing support and DNS > update support, without adding files from external projects. > > The drawback is that libbind appears to be packaged only as a static > library on Debian based systems, whereas libresolv is available as a > shared library on virtually every system. libbind.a is 428694 bytes > in size on my Kubuntu system and 546908 bytes in size on my Fedora > system. > > -Hans > > Changelog > Switch from libresolv to libbind for DNS support. The disadvantage is that bind-libs is under a BSD 4 clause license. While it is allowed to link it, it will also make dnsapi and likely WINE itself fall under the BSD 4 clause license (additionaly to LGPL) and both need to be honored from this point on. BSD 4 clause is only used in parts of the lib, other parts are under regular BSD license or others. grep -r "All advertising materials" <dir> to find all files with BSD clause 4 license. I think the DNS parsing code itself in libbind is under normal BSD license and the pieces necessary could be copied/imported safely into LGPL code. (This is likely the reason Debian packs it statically.) Ciao, Marcus

2 1

Re: Dinput8: DirectInput8Create rewrite
by Jesse Allen 15 May '06

15 May '06

On 5/14/06, Roderick Colenbrander <thunderbird2k(a)gmx.net> wrote: > Hi, > > This patch lets DirectInput8Create work similar to the native version which > according to MSDN uses CoCreateInstance to create a dinput8 object. This > rewrite should fix the dll count related issues. > This patch needs linking against ole32 and it breaks keyboard and mouse input in Star Wars Battlefront for some reason. Hmmm... Jesse

1 0

Moving menus to wineserver
by Dan Kegel 15 May '06

15 May '06

wineserver currently tracks a lot of info about windows (those things on the screen, not the operating system). It turns out (see http://bugs.winehq.org/show_bug.cgi?id=5231 ) that it probably needs to do the same thing for menus and menuitems to make the win32 apis GetMenuItemCount and GetMenuItemInfo work for menus of other processes. (Why would you need that, you say? To support gui test scripting languages, e.g. Perl's win32::GuiTest; see http://kegel.com/wine/testsw.html I have a feeling that big companies won't take Wine seriously until they can run their apps' regression tests on Wine. So this is, if not urgent, at least pretty desirable.) This seems to be part of Dimi Paun's TODO list. Dimi, were you about to do this? If not, am I on the right track here? Thanks, Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

2 3

Still missing glyph
by Uwe Bonnes 14 May '06

14 May '06

Hallo, even with a recent fontforge, I get "Missing glyph for": fontforge -script ../fonts/genttf.ff small_fonts.sfd small_fonts.ttf Copyright (c) 2000-2006 by George Williams. Executable based on sources from 19:18 13-Apr-2006. ../tools/widl/widl -I. -I. -I../include -I../include -h -H amstream.h amstream.idl LD_LIBRARY_PATH="../libs/unicode:$LD_LIBRARY_PATH" ../tools/sfnt2fnt system.ttf 16 932 96 128 7 LD_LIBRARY_PATH="../libs/unicode:$LD_LIBRARY_PATH" ../tools/sfnt2fnt small_fonts.ttf 11 1252 96 128 5 LD_LIBRARY_PATH="../libs/unicode:$LD_LIBRARY_PATH" ../tools/sfnt2fnt small_fonts.ttf 11 1250 96 128 5 LD_LIBRARY_PATH="../libs/unicode:$LD_LIBRARY_PATH" ../tools/sfnt2fnt small_fonts.ttf 11 1253 96 128 5 Missing glyph for char 0385 Missing glyph for char 0386 ... Is this expected behaviour? -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

2 1

Valgrind warning cleanup...
by Dan Kegel 14 May '06

14 May '06

I was trying to use valgrind-3.1.1 to track down an OpenOffice crash, and I noticed that current wine shows errors like this even running notepad: ==8986== Invalid read of size 4 ==8986== at 0x20010C21: (within /lib/ld-2.3.6.so) ... ==8986== by 0x20189D10: dlopen (in /lib/tls/i686/cmov/libdl-2.3.6.so) ==8986== by 0x20023CB8: wine_dlopen (loader.c:663) ==8986== by 0x200248B7: wine_init (loader.c:625) ==8986== by 0x7BF00F9F: main (main.c:58) ==8986== Address 0x2018D3A8 is 56 bytes inside a block of size 57 alloc'd ==8986== at 0x2001C422: malloc (vg_replace_malloc.c:149) ==8986== by 0x2002463A: first_dll_path (loader.c:201) ==8986== by 0x20024895: wine_init (loader.c:623) ==8986== by 0x7BF00F9F: main (main.c:58) I looked at first_dll_path, but I couldn't see any problem. On the theory that glibc is reading a bit past the end of the string, I tried changing the line that does the allocation in question. Adding one byte cut down the number of those warnings; adding two bytes didn't help more than adding one; adding three bytes got rid of the warnings. Here's the change: --- libs/wine/loader.c 17 Mar 2006 12:23:52 -0000 1.36 +++ libs/wine/loader.c 13 May 2006 19:56:52 -0000 @@ -199,7 +199,7 @@ char *p; int namelen = strlen( name ); - context->buffer = malloc( dll_path_maxlen + 2 * namelen + strlen(ext) + 3 ); + context->buffer = malloc( dll_path_maxlen + 2 * namelen + strlen(ext) + 6 ); context->index = build_dir ? 0 : 3; /* if no build dir skip all the build dir magic cases */ context->name = context->buffer + dll_path_maxlen + namelen + 1; context->namelen = namelen + 1; What do folks think - should we just make the allocation a little bigger to make valgrind happy? I'd also like to understand what's causing them, too, though. - Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

3 5

Re: widl: Added an option to set importlib search path.
by Detlef Riekenberg 14 May '06

14 May '06

Hi Jacek: +" -L path Set importlib searcg dir to path (multiple -L Small typo in tools/widl/widl.c: "search" -- By By ... ... Detlef

1 0