wine-devel

wine-devel@winehq.org

11 participants
84461 discussions

Re: Directx9
by Ivan Leo Puoti 13 Sep '04

13 Sep '04

> If we as a project are committed to doing 0.9, 1.0+ releases it means we > can't simply break code people are using to play their games for months > at a time and shrug it off After 1.0 there will be a stable and a development branch, the stable one for users, and the development one for new stuff, where it doesn't matter if things are broken for a while. A bit like the kernel with 2.4 stable and 2.5 full of broken/buggy stuff until 2.6 is ready, and so on. At least that's how most projects work. Ivan.

3 2

Re: Http Fixes
by Marcus Meissner 13 Sep '04

13 Sep '04

On Mon, Sep 13, 2004 at 11:38:32AM +0100, Robert Shearman wrote: > Hi, > > I have seen one server that likes to return "ICY" instead of "HTTP/1.x". > This patch allows the connection to the server continue, as native does. This would be "icecast" I suspect ;) Ciao, Marcus

1 0

Re: Wininet: InternetSetStatusCallback Fix
by Robert Shearman 13 Sep '04

13 Sep '04

Robert Shearman wrote: > Hi, > > You can call InternetSetStatusCallback on a HTTP handle, not just a > InternetOpen handle, so we should search up through the parents to > find it. Oops, I just checked on MSDN and this patch is wrong. We should copy and store the status callback in every handle, not just in the top level one. I'll submit a new patch soon. Rob

1 0

PEB documentation
by Mike Hearn 13 Sep '04

13 Sep '04

This is interesting information, but where did it come from? Is this structure described in some MS documentation? If so, does it have the full layout of the TEB as well? > Log message: > Robert Reif <reif(a)earthlink.net> > Document all the structure members up to SessionId in the PEB.

1 0

Re: winecfg (was Re: W->A calls)
by Mike Hearn 13 Sep '04

13 Sep '04

> It appears that Mike has been doing a lot of work on this. > http://www.winehq.org/hypermail/wine-devel/2004/09/0221.html > So you probably first want to coordinate with him. Yes. I'm aware that I'm holding people up right now due to my non-connectedness. I'm actually living in a bed&breakfast with no net connection, so I only have web/email access at work. It's kind of odd - I've been doing some work on it lately, and suddenly people pop up who want to work on it after a long period of inactivity. Oh well :) > I would start by running "winecfg", and seeing what doesn't work, which > is a lot. For example, on the very first tab, you can select a "Windows > Version", but changing the setting doesn't actually change anything (it > should make/alter a registry entry). The winecfg code is in > wine/programs/winecfg. Fixed by me already. Actually, I've done quite a bit of work on it, I'd recommend waiting for 2 weeks at which point I'll have net access to my laptop again and can hopefully send off my work. At that point I'll be disappearing back into university life and doing CW work again so the coast will be clear for others to work on it. > A long discussion of things that are needed is in the thread starting: > http://www.winehq.org/hypermail/wine-devel/2004/08/0007.html I've sorted some of these already, especially in the apps tab. So ... if you guys can wait for 2 weeks I should (hopefully) have something to show soon. thanks -mike

1 0

RFC: ntdll critical section debug
by Robert Reif 13 Sep '04

13 Sep '04

I have been running into deadlock problems in the direct sound dll and have been playing with the debug part of critical sections. I added a lock name to Spare[1] to give the critical sections names that will show up in the error messages just like wine's ntdll does. After looking at the ntdll code, it looks like this causes a leak because ntdll statically initializes it's locks and doesn't delete the debug part when it finds Spare[1] used. I also tried adding a linked list of critical section debug structs just like real windows does. Because ntdll statically initializes it's locks, they don't get added to the list. I manually stuff the critical section list lock into the list so it shows up and user locks show up but not the other internal ntdll locks. Is there a good reason to statically initialize ntdll locks or can they be initialized like normal locks? Here is a patch that adds the linked list of debug info: Index: dlls/ntdll/critsection.c =================================================================== RCS file: /home/wine/wine/dlls/ntdll/critsection.c,v retrieving revision 1.27 diff -u -r1.27 critsection.c --- dlls/ntdll/critsection.c 6 Sep 2004 20:26:23 -0000 1.27 +++ dlls/ntdll/critsection.c 10 Sep 2004 16:29:14 -0000 @@ -36,6 +36,51 @@ WINE_DEFAULT_DEBUG_CHANNEL(ntdll); WINE_DECLARE_DEBUG_CHANNEL(relay); +static LIST_ENTRY RtlCriticalSectionList; +static CRITICAL_SECTION RtlCriticalSectionLock; +static CRITICAL_SECTION_DEBUG RtlCriticalSectionDebug = +{ + 0, 0, &RtlCriticalSectionLock, + { &RtlCriticalSectionDebug.ProcessLocksList, + &RtlCriticalSectionDebug.ProcessLocksList }, + 0, 0, { 0, (DWORD)(__FILE__ ": RtlCriticalSectionLock") } +}; +static CRITICAL_SECTION RtlCriticalSectionLock = { &RtlCriticalSectionDebug, -1, 0, 0, 0, 0 }; + +void critsection_init() +{ + InitializeListHead( &RtlCriticalSectionList ); + InsertHeadList( &RtlCriticalSectionList, &RtlCriticalSectionDebug.ProcessLocksList ); +} + +static void CRIT_Dump() +{ + int i = 0; + PLIST_ENTRY mark, entry; + PCRITICAL_SECTION_DEBUG debug; + + RtlEnterCriticalSection( &RtlCriticalSectionLock ); + + mark = &RtlCriticalSectionList; + for (entry = mark->Flink; entry != mark; entry = entry->Flink) + { + debug = CONTAINING_RECORD(entry, CRITICAL_SECTION_DEBUG, ProcessLocksList); + + TRACE("RtlCriticalSectionDebug[%d]:\n", i++); + TRACE(" Type = %04x\n", debug->Type); + TRACE(" CreatorBackTraceIndex = %04x\n", debug->CreatorBackTraceIndex); + TRACE(" CriticalSection = %p\n", debug->CriticalSection); + TRACE(" ProcessLocksList.Flink = %p\n", debug->ProcessLocksList.Flink); + TRACE(" ProcessLocksList.Blink = %p\n", debug->ProcessLocksList.Blink); + TRACE(" EntryCount = %ld\n", debug->EntryCount); + TRACE(" ContentionCount = %ld\n", debug->ContentionCount); + TRACE(" Spare[0] = %08lx\n", debug->Spare[0]); + TRACE(" Spare[1] = %08lx (%s)\n", debug->Spare[1], debugstr_a((char *)debug->Spare[1])); + } + + RtlLeaveCriticalSection( &RtlCriticalSectionLock ); +} + inline static LONG interlocked_inc( PLONG dest ) { return interlocked_xchg_add( dest, 1 ) + 1; @@ -126,12 +171,16 @@ crit->DebugInfo->Type = 0; crit->DebugInfo->CreatorBackTraceIndex = 0; crit->DebugInfo->CriticalSection = crit; - crit->DebugInfo->ProcessLocksList.Blink = &(crit->DebugInfo->ProcessLocksList); - crit->DebugInfo->ProcessLocksList.Flink = &(crit->DebugInfo->ProcessLocksList); crit->DebugInfo->EntryCount = 0; crit->DebugInfo->ContentionCount = 0; crit->DebugInfo->Spare[0] = 0; crit->DebugInfo->Spare[1] = 0; + RtlEnterCriticalSection( &RtlCriticalSectionLock ); + InsertTailList( &RtlCriticalSectionList, &(crit->DebugInfo->ProcessLocksList) ); + RtlLeaveCriticalSection( &RtlCriticalSectionLock ); + + if (TRACE_ON(ntdll)) + CRIT_Dump(); } } crit->LockCount = -1; @@ -195,12 +244,19 @@ crit->LockSemaphore = 0; if (crit->DebugInfo) { + RtlEnterCriticalSection( &RtlCriticalSectionLock ); + RemoveEntryList( &(crit->DebugInfo->ProcessLocksList) ); + RtlLeaveCriticalSection( &RtlCriticalSectionLock ); + /* only free the ones we made in here */ if (!crit->DebugInfo->Spare[1]) { RtlFreeHeap( GetProcessHeap(), 0, crit->DebugInfo ); crit->DebugInfo = NULL; } + + if (TRACE_ON(ntdll)) + CRIT_Dump(); } return STATUS_SUCCESS; } Index: dlls/ntdll/ntdll_misc.h =================================================================== RCS file: /home/wine/wine/dlls/ntdll/ntdll_misc.h,v retrieving revision 1.48 diff -u -r1.48 ntdll_misc.h --- dlls/ntdll/ntdll_misc.h 15 Jul 2004 22:07:21 -0000 1.48 +++ dlls/ntdll/ntdll_misc.h 10 Sep 2004 16:29:14 -0000 @@ -50,6 +50,7 @@ extern void debug_init(void); extern void thread_init(void); extern void virtual_init(void); +extern void critsection_init(void); /* server support */ extern void server_init_process(void); Index: dlls/ntdll/thread.c =================================================================== RCS file: /home/wine/wine/dlls/ntdll/thread.c,v retrieving revision 1.21 diff -u -r1.21 thread.c --- dlls/ntdll/thread.c 23 Aug 2004 18:52:54 -0000 1.21 +++ dlls/ntdll/thread.c 10 Sep 2004 16:29:14 -0000 @@ -134,6 +134,7 @@ debug_info.str_pos = debug_info.strings; debug_info.out_pos = debug_info.output; + critsection_init(); debug_init(); virtual_init();

3 2

spying COM
by Saulius Krasuckas 13 Sep '04

13 Sep '04

Am not sure would it be of any use for list members, but just some minutes ago I've discovered "The COM spy". Quite old stuff, but whatever: http://staff.develop.com/jasonw/comspy/default.htm

2 1

epoll patch - status?
by Shachar Shemesh 12 Sep '04

12 Sep '04

Hi Alexandre, I'm sorry for being a pest about this. I was wondering whether there is anything that needs to be done on that one in order to facilitate committing this patch? Shachar -- Shachar Shemesh Lingnu Open Source Consulting ltd. http://www.lingnu.com/

5 14

Re: IShellFolder::CompareIDs partially undocumented take 2
by Vitaliy Margolen 12 Sep '04

12 Sep '04

Is there are any problems with this patch? Tests do confirm it. Vitaliy Margolen Saturday, September 4, 2004, 9:48:41 PM, Vitaliy Margolen wrote: > Since my previous patch wasn't excepted here it is again. This time with more fixes > and tests. > Also it _is_ documented that CompareIDs returns result in lower 16 bits. And this function > returns only -1/0/1 as a result. > Vitaliy Margolen > changelog: > dlls/shell32/shlfolder.c > IShellFolder::CompareIDs - return only -1/0/1 in lower 16 bit.

1 0

W->A calls
by Joris Huizer 11 Sep '04

11 Sep '04

Hello, By accident, I removed the emails of the thread I started, but I'll look it up if necessary; I found this old wine issue of some time ago.. I think the patch it has there could be helpfull http://winehq.org/?issue=229#About%20Converting%20W-%3EA%20Calls However, I don't know how to make simple tests.. as I'd really need to test wether the code I write works, and works properly :p How do I write tests for that? I'm running linux (in practice, I never boot into windows) so how should I do this (testing without windows environment)? How are tests supposed to run? I hope you can give some clues on that; I found there are many W->A calls in dlls/winmm/winmm.c (according to the janitorial page) so I might start there (hoping not all of them are so tricky ;) ) Thanks for suggestions on this, Joris

7 9

Re: Correct ConvertSidToStringSidW with test
by Dimitrie O. Paun 11 Sep '04

11 Sep '04

On Sat, Sep 11, 2004 at 08:37:09AM -0700, Juan Lang wrote: > ConvertSidToStringSidW was incorrect for low-value > authorities; it put spaces in the resulting string. > This corrects that, and includes a test case for it. Content-Description: advapi32.diff Uniffied diff, please :) -- Dimi.

1 0

Re: PROPSHEET: Wizard resizing
by Filip Navara 11 Sep '04

11 Sep '04

Filip Navara wrote: > I know this patch isn't really correct, but the old code isn't correct > either. Without it about 80% of the wizards were resized incorrectly > (or not at all) and the header line wasn't hidden sometimes. Even > though I would like to code the resizing correctly it will require > major code shuffling that I can't do now due to time constraints. So I > believe for now this patch is less evil than letting the code as is. > > (I can provide more details about how it should be implemented > correctly on request) > > - Filip > > Changelog: > First attempt at correcting wizard resizing. Scratch this patch please. I'll try to implement it correctly... Regards, Filip

1 0

Wine and non ANSI charset fonts
by Andreas Rosenberg 11 Sep '04

11 Sep '04

Hello guys, after a long pause I'm back testing our application (http://appdb.winehq.org/appview.php?appId=1536) with WINE. In the meantime I've been working to prepare our application for the east european/asian market (UTF16, wide char API). This seems to work so far, but as soon as you enter text requiring non ANSI glyphs, I'm only getting "square boxes" instead of the desired characters. So I was searching for another application that uses non ANSI glyhps with WINE. I choosed Opera, because it's quite easy to install. But unfortunately Opera 7.x crashed after entering the first URL. The function "WSAStringToAddress" and some other WSA funtions were missing/stubs. Yesterday evening I fixed these problems and just sent a patch making Opera work with Wine. With Opera I'm now getting the same or similar effects. just try the Opera in Wine and go to: http://www.viatech.com.tw/big5/index/index.jsp So my question: is this a configuration problem? Konquerer displays the page correctly, so the fonts on the Linux side have the correct glyphs available. But it seems they are not available in WINE - I looked into the WINE config file and found something like this: ;;default TrueType fonts with russian koi8-r encoding "Default" = "-monotype-arial-*-*-*--*-*-*-*-*-*-koi8-r" "DefaultFixed" = "-monotype-courier new-*-*-*--*-*-*-*-*-*-koi8-r" "DefaultSerif" = "-monotype-times new roman-*-*-*--*-*-*-*-*-*-koi8-r" "DefaultSansSerif" = "-monotype-arial-*-*-*--*-*-*-*-*-*-koi8-r" This seems to define something regarding the cyrilic charset. With these lines enabled, Opera can display russian pages correctly. But what do I need to enter to get the glyphs of other character sets? (chinese, hangul, etc...) I read the Wine users guide regarding fonts, but it still does not address the character set topic.Can anybody shed some light on this? Regards Andy

2 1

Re: Implementation of WSAStringToAddress(A/W), WSAHtonl, WSAHtons
by Alexandre Julliard 10 Sep '04

10 Sep '04

Andreas Rosenberg <andreas_rosenberg(a)web.de> writes: > +INT WINAPI WSAStringToAddressW(LPCWSTR AddressString, > + INT AddressFamily, > + LPWSAPROTOCOL_INFOA lpProtocolInfo, You should have an LPWSAPROTOCOL_INFOW here instead, and convert it before calling WSAStringToAddressA. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: Almost complete UpdateResource implementation
by Alexandre Julliard 10 Sep '04

10 Sep '04

Vincent Béron <vberon(a)mecano.gme.usherb.ca> writes: > + if(!updates_list) > + { > + updates_list = current_updates; > + list_init(&updates_list->entry); > + } > + else > + list_add_after(&updates_list->entry, &current_updates->entry); That's not the right way to use the list functions. updates_list should be a separate struct list, that will make things a lot easier. > + LIST_FOR_EACH(ptr, &current_updates->resources_list->entry) > + { > + current_resource = LIST_ENTRY(ptr, QUEUEDRESOURCE, entry); > + free_queuedresource(current_resource); > + } You cannot free objects that are still on list, you have to remove them first. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: CreateDirectoryA conformance test improvement
by Alexandre Julliard 10 Sep '04

10 Sep '04

Wolfgang Schwotzer <wolfgang.schwotzer(a)gmx.net> writes: > This patch provides a conformance test for my patch > sent on 06. Sept 2004. > > I have successfully tested it on Windows98 and WindowsNT4. Actually the results of that test depend on what the current directory is on the drive you are testing. If your current directory is \ you will see that it fails on Windows too. I believe the Wine code is correct, it returns ACCESS_DENIED on the root and ALREADY_EXISTS anywhere else, just like Windows. -- Alexandre Julliard julliard(a)winehq.org

1 0

COmpilation problems with older Freetype 2
by Vincent Béron 10 Sep '04

10 Sep '04

A patch by Huw broke compilation of Wine on RH8 (and probably other platforms as well) which use Freetype 2.1.3 or earlier. More specifically, FT_ENCODING_NONE is undefined. Other ft_encoding_* (or FT_ENCODING_*) problems have arised in the past, but the solution used identifiers already present in older Freetype versions. The attached patch works (and results in the same compiled code IMHO), but looks like a hack rather than a proper fix (use MS_MAKE_TAG? that has already been removed in the past to use ft_encoding_*). Vincent

1 0

Re: Fix twain device handle corruption
by Alexandre Julliard 10 Sep '04

10 Sep '04

Francois Gouget <fgouget(a)codeweavers.com> writes: > * or we could check that config.h has been included first using > something like: > > #ifndef __WINE_CONFIG_H > # error You must include config.h first > #endif I think that's preferable, we really want to be using the same structure everywhere. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [SHLWAPI] Implemented PathSearchAndQualify (another fix)
by Vincent Béron 10 Sep '04

10 Sep '04

Le ven 10/09/2004 à 13:22, Diego 'Flameeyes' Pettenò a écrit : > I have fixed the patch for the problems Rolf Kalbermatter pointed out. Hope > this time is the good one :) Not likely if the patch is not attached... ;) Vincent

2 1

Re: winedos:MCB implementation
by Andreas Mohr 10 Sep '04

10 Sep '04

Hi, On Fri, Sep 10, 2004 at 06:09:26PM +0200, Markus Amsler wrote: > That's some of the result of my effort getting FreeCOM running in winedos. > It's quite important to winedos having proper MCB support, a lot of apps > depend on it. Indeed! Great, thanks a lot for this very important work!! > It also enables running/terminating apps within dos shells, and opens > the door to TSR. > With some more patches (IO-redirection, TSR comming soon) i'm able to > run FreeCOM, compile FreeCOM within FreeCOM, run XRAY and getting my TSR > app starting. Cool. > Please let me know, if there's something wrong with the patch (design > problems, naming conventions, formating, ...). Looks nice and clean to me on cursory glance. Andreas Mohr

1 0

Re: dinput axis mapping and format mapping patch
by Alexandre Julliard 10 Sep '04

10 Sep '04

Robert Reif <reif(a)earthlink.net> writes: > Add a [dinput] section to your config file and provide a line with the > following information: > > "your device name" = "X,Y,Z,Rx,Ry,Rz,Slider1,Slider2,POV1,POV2,POV3,POV4" > > where "your device name" is what you get from /bin/jstest and > the axes list are the axes supported by your joystick. > > For example, here is part of my config file: > > [dinput] > "Logitech Logitech Extreme 3D Pro" = "X,Y,Rz,Slider1,POV1" You should add this in the registry instead of the config file, this way we won't have to move it later on. HKLM/Software/Wine/Dinput should be a good place for it. -- Alexandre Julliard julliard(a)winehq.org

3 6

Fwd: Cron <paulm@quisquiliae> $HOME/Production/cronCrossBuild 10:00
by Paul Millar 10 Sep '04

10 Sep '04

Hi everyone, A couple of quick things: First, I'm off on holiday for a week (starting tomorrow :^) so won't be able to answer emails for a bit. The cross-building should carry on regardless, though. Second, very occasionally get messages like: ---------- Forwarded Message ---------- Subject: Cron <paulm@quisquiliae> $HOME/Production/cronCrossBuild 10:00 Date: Friday 10 September 2004 11:18 From: root of quisquiliae <p.millar(a)physics.gla.ac.uk> To: p.millar(a)physics.gla.ac.uk cvs checkout: warning: unrecognized response `cvs: buffer.c:1384: stdio_buffer_shutdown: Assertion `fstat ( fileno (bc->fp), &s ) != -1 failed.' from cvs server cvs [checkout aborted]: end of file from server (consult above messages if any) ------------------------------------------------------- which looks like WineHQ's cvs server sulking. Does any else get these? Cheers, Paul.

1 0

Linking with winelib
by Nikolay A. Liber 10 Sep '04

10 Sep '04

Hello I am using winelib with Mono hack to load windows DLL into python module. There is a dummy function wine_pthread_init_thread in libwine.so. wine_pthread_init_thread implemented in winelib.so.exe from mono shared winelib. I link module. gcc -shared mymodule.so module_code.o -lwine ./winelib.so.exe But when I load module init_thread from ntdll calls dummy wine_pthread_init_thread. How to override that dumy function? Thank you Nikolay

2 1

RE: modal dialog: Wait until message queue runs dry before calling ShowWindow()
by Zach Gorman 09 Sep '04

09 Sep '04

Yes, the test code is on its way. These patches are all from a few weeks ago, but Alexandre asked me to break them up into multiple smaller patches. I tested them with the latest tree and they seem to still work fine. Zach -----Original Message----- From: Mike McCormack [mailto:mike@codeweavers.com] Sent: Thursday, September 09, 2004 2:37 PM To: zach(a)archetypeauction.com Subject: Re: modal dialog: Wait until message queue runs dry before calling ShowWindow() Zach Gorman wrote: > My tests show that Windows behaves this way. If you could submit your test as a patch including it in the wine test suite, that would be great. If not, just posting it with a LGPL compatible license so others can adapt it to the test suite would be appreciated too. Mike

1 0

Re: edit control
by Robert Shearman 09 Sep '04

09 Sep '04

Ulrich Czekalla wrote: >ChangeLog: > Ulrich Czekalla <ulrich(a)codeweavers.com> > Allow copying selection when control is read only > > >------------------------------------------------------------------------ > >Index: dlls/user/edit.c >=================================================================== >RCS file: /home/wine/wine/dlls/user/edit.c,v >retrieving revision 1.1 >diff -u -r1.1 edit.c >--- dlls/user/edit.c 31 Aug 2004 01:10:08 -0000 1.1 >+++ dlls/user/edit.c 9 Sep 2004 13:01:26 -0000 >@@ -3604,7 +3604,8 @@ > > /* Protect read-only edit control from modification */ > if(es->style & ES_READONLY) >- return; >+ if (c != 0x03) /* Allow copying selection */ >+ return; > > control = GetKeyState(VK_CONTROL) & 0x8000; > > That read-only check looks completely bogus to me as the code below already checks for ES_READONLY. Rob

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

wine-devel