http://bugs.winehq.org/show_bug.cgi?id=10584
Summary: Last developments of Wine broke the setup of Trados Product: Wine Version: CVS/GIT Platform: HP OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: wine-misc AssignedTo: wine-bugs@winehq.org ReportedBy: yolande@haneder.biz
Created an attachment (id=9387) --> (http://bugs.winehq.org/attachment.cgi?id=9387) log file without relay
You can find attached the short version. The only difference towards the former logs are the ole errors (new)
If you need, I can send you a log with +relay. It's quite big, more than 1 GB and 84 MB as zip file. It may be too big to know where the error come from.
If you wish I can make another log.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #1 from Stefan Dösinger stefandoesinger@gmx.at 2007-11-28 12:21:55 --- Can you do a regression test?
http://wiki.winehq.org/RegressionTesting
http://bugs.winehq.org/show_bug.cgi?id=10584
Yolande Haneder yolande@haneder.biz changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution| |ABANDONED
--- Comment #2 from Yolande Haneder yolande@haneder.biz 2007-11-29 01:35:51 --- Sorry I can't - It's better to close the bug. To make a regression test, I would need someone to compile me each GIT version, which is not to be expected.
NET 1.1 and 2 are affected too so it will be cleared sometime. (following log is debug+all for net 1.1 since the file for Trados is too big) www.uebersetzen.at/log2.txt.zip
http://bugs.winehq.org/show_bug.cgi?id=10584
Kai Blin blin@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |UNCONFIRMED Resolution|ABANDONED |
--- Comment #3 from Kai Blin blin@gmx.net 2007-11-29 01:40:24 --- Is there a demo version of your program that shows similar problems?
Reopening bug.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #4 from Yolande Haneder yolande@haneder.biz 2007-11-29 01:49:14 --- It's better that you try to look on the NET. It's broken too.
SDL removed the demo from their website, each version can be a demo so SDL could send you a demo if you want - just have to ask them.
One things that came up to my mind to get that far: I have *all* windows font installed on my linux box (I did it to get all Windows fonts to work on Open Office for a better compatibility). It may be needed to get that far.
Anyway, installing Trados is not an easy thing, we may need to contact through some instant messenger for a step by step instruction (I am doing it blindly now so I may not explain everything at the first time).
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #5 from Yolande Haneder yolande@haneder.biz 2007-11-29 02:09:04 --- You will need to contact SDL anyway for a demo version. It is build 820 (8.0.0.820) professional edition which is very limited because it was only for the first orders and had been replaced by build 826 a month later. The new version, SP1 has a new windows installer creating me problems on my old XP so I can not guarantee that it will be installed without problems (I have it to test it someday if I have time).
http://bugs.winehq.org/show_bug.cgi?id=10584
Yolande Haneder yolande@haneder.biz changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #9387 is|0 |1 obsolete| |
--- Comment #6 from Yolande Haneder yolande@haneder.biz 2007-12-07 05:17:56 --- Created an attachment (id=9532) --> (http://bugs.winehq.org/attachment.cgi?id=9532) Updated log
Update with a clean install on the GIT of yesterday (06.12).
First of all, I want to thank you because an ole error had turned into a fixme and it seems that something is happening.
However the same sequence is happening again and again. Is it normal? (see attachment).
The setup exception blocking everything is still there.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #7 from Yolande Haneder yolande@haneder.biz 2007-12-08 00:00:07 --- With the GIT of 7.12 some improvements:
.NET 2 is installing again (as far as this issue is concerned). However the behavior of last log is still there.
Summing up: .NET 2 does not have a setup exception anymore, Trados still does.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #8 from Yolande Haneder yolande@haneder.biz 2007-12-12 02:43:25 --- The problem had now been shifted so I made a new log with +ole since most if not all of the errors before the setup abort had been with ole.
You can find the log under www.uebersetzen.at/TradosSetupError.txt.zip.
Even as a zip it is still much too big to attach it here.
I add the following error because it is recurring quite a lot in the installation, it may or may not be related.
trace:ole:apartment_getclassobject calling ole32!DllGetClassObject trace:ole:PSFacBuf_CreateStub ({00000010-0000-0000-c000-000000000046},0x530b50,0x7d19a280) fixme:ole:PSFacBuf_CreateStub stubbing not implemented for ({00000010-0000-0000-c000-000000000046}) yet! err:ole:marshal_object Failed to create an IRpcStubBuffer from IPSFactory for {00000010-0000-0000-c000-000000000046} with error 0x80004005
The memory leak is now at: err:ntdll:RtlpWaitForCriticalSection section 0x7bc8c644 "loader.c: loader_section" wait timed out in thread 0046, blocked by 001b, retrying (60 sec)
http://bugs.winehq.org/show_bug.cgi?id=10584
Dan Kegel dank@kegel.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |dank@kegel.com
--- Comment #9 from Dan Kegel dank@kegel.com 2007-12-12 08:33:55 --- For what it's worth, I've had very good luck compressing huge log files with rzip. It's amazingly fast and compresses incredibly well.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #10 from Yolande Haneder yolande@haneder.biz 2007-12-15 01:12:23 --- An Update with 0.9.51:
The memory leak at the end had been fixed. I am pasting the most relevant part instead of having to send a new log with over 100 MB (I have not yet found how to use rzip) Now the log is terminating with
trace:ole:apartment_construct creating new apartment, model=0 trace:ole:apartment_construct Created apartment on OXID 980000cafe trace:ole:CoInitializeEx ((nil), 0) trace:ole:apartment_get_or_create entering the multithreaded apartment 980000cafe trace:ole:apartment_addref 980000cafe: before = 1 err:seh:setup_exception nested exception on signal stack in thread 0099 eip 7bc622b9 esp 7ffdbbf0 stack 0x241000-0x350000 trace:ole:DllMain 0x7e900000 0x3 (nil) trace:ole:DllMain 0x7e900000 0x3 (nil) trace:ole:DllMain 0x7ecd0000 0x3 (nil) trace:ole:DllMain 0x7ecd0000 0x3 (nil) trace:ole:DllMain 0x7e900000 0x3 (nil) trace:ole:DllMain 0x7ecd0000 0x3 (nil)
Another added error
trace:ole:CoGetStandardMarshal ({00020401-0000-0000-c000-000000000046},0x1c3838,0,(nil),0,0x7cc99824) trace:ole:CoMarshalInterface Using standard marshaling err:ole:CoMarshalInterface Failed to write OBJREF header to stream, 0x80030070 trace:ole:CoMarshalInterface completed with hr 0x80030070 trace:ole:RpcStream_Release size=0 warn:ole:CStdStubBuffer_Invoke a stub call failed with exception 0x80030070 (-2147286928) trace:ole:TMStubImpl_Release (0x8925f8)->(ref before=2) trace:ole:apartment_release 210000001f: after = 1 err:rpc:I_RpcReceive we got fault packet with status 0x80030070
Otherwise it is pretty much the same.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #11 from Yolande Haneder yolande@haneder.biz 2007-12-20 01:29:52 --- The above problem had now been fixed, a new one started.
Trados is aborting with the message that the setup had been interrupted and the system had not been modified. To complete the installation, I should run the setup another time (which is obviously failing).
On the log, I marked what belonged to trados or not to make things a bit clearer. This time it seems to be a msi error. Please advise me if you wish details on a particular point.
http://bugs.winehq.org/show_bug.cgi?id=10584
Yolande Haneder yolande@haneder.biz changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #9532 is|0 |1 obsolete| |
--- Comment #12 from Yolande Haneder yolande@haneder.biz 2007-12-20 01:34:38 --- Created an attachment (id=9713) --> (http://bugs.winehq.org/attachment.cgi?id=9713) New log again
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #13 from Yolande Haneder yolande@haneder.biz 2007-12-21 01:38:18 --- Back and forth...
Yesterday with the GIT of 19th, I had the Installshield crashing but the lines fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a99c,0x7cc8a9a0), stub! fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a998,0x7cc8a99c), stub! fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a954,0x7cc8a958), stub! fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a99c,0x7cc8a9a0), stub!
appeared only once. Today, with the GIT of 20th, the installshield error had been fixed but the rpc is calling the above lines again and again like 2 weeks ago (of course now without the error in between which had been fixed.
Now the ending lines are:
fixme:virtual:NtAllocateVirtualMemory MEM_WRITE_WATCH type not supported err:ole:CoGetObjectContext apartment not initialised err:seh:setup_exception nested exception on signal stack in thread 00bc eip 7bc622b9 esp 7ffdbbf0 stack 0x241000-0x350000 fer_GetDestCtx (0x7cc8a99c,0x7cc8a9a0), stub! bash: syntax error near unexpected token `0x7cc8a99c,0x7cc8a9a0' yolande@linux-fsf0:~/.wine/drive_c/Yolande_Privat> fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a998,0x7cc8a99c), stub! bash: syntax error near unexpected token `0x7cc8a998,0x7cc8a99c' yolande@linux-fsf0:~/.wine/drive_c/Yolande_Privat> fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a954,0x7cc8a958), stub! bash: syntax error near unexpected token `0x7cc8a954,0x7cc8a958' yolande@linux-fsf0:~/.wine/drive_c/Yolande_Privat> fixme:ole:ServerRpcChannelBuffer_GetDestCtx (0x7cc8a99c,0x7cc8a9a0), stub! bash: syntax error near unexpected token `0x7cc8a99c,0x7cc8a9a0'
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #14 from Yolande Haneder yolande@haneder.biz 2007-12-29 11:44:09 --- Result of the regression test affecting the setup error; I have no idea how everything hangs together
8d571e28f1779c82ba91973dc540d0d7ec5ef867 is first bad commit commit 8d571e28f1779c82ba91973dc540d0d7ec5ef867 Author: Paul Millar paul@astro.gla.ac.uk Date: Fri Nov 9 18:07:36 2007 +0100
libwine: Fix naked getuid() to allow compilation under MinGW.
:040000 040000 ee36bfdcf5d65ec8bb1c8ee478e74e0af8679039 9429fc53e7b3108987c33bfef19ff0e710d1f2e3 M libs
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #15 from Yolande Haneder yolande@haneder.biz 2007-12-29 13:23:15 --- The only thing I can think of is that this patch is creating an interference between the installation of Trados and SDLX (other components are not affected)and the Linux Java Runtime Environment while building the Java classes (since the Linux Java is acting before the Windows Java).
Furthermore since the setup error there is no interaction anymore between the diverse parts. Maybe it would be interesting to try with another Windows application using a Linux Java to install.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #16 from Dan Kegel dank@kegel.com 2007-12-29 13:57:58 --- If the installer uses java, have you tried installing a windows java runtime before installing?
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #17 from Yolande Haneder yolande@haneder.biz 2007-12-30 00:56:52 --- The Windows Java is part of the installation itself (before the installation of Trados). Seen that way, I have two Java (Windows and Linux) running at the same time, whereby however the java classes are always saved in the Linux temp folder. The Linux Java is however already there when installing the Windows Java. Another test is making me think, Java for Linux is at work:
When the installation is hanging and I am closing the console window, it is still hanging. Open another console window and make wineserver -k, it is still running/hanging.
Making wineserver -k in the same console window where I launched the installation and Trados is killed.
Java for Windows is recently having problems to install (since this wine version) so it is the matter for another regression test.
Some notes about regression test: The description of the path to a regression test is not working for X86_64 suse computer (may work for unbuntu?)
It is written for a 64 bit processor to run CC"ccache gcc -m32" ./configure.
This does not work for a X86_64 distribution because it is still picking the x86_64 libraries. The configure will be OK but it will crash while compiling.
As seen elsewhere, to compile Wine on a x86_64 linux SUSE you have to enter: ./ configure --x-libraries=/usr/X11R6/lib In this way it is also working as with "gcc -m32" (as seen while compiling)
Entering CC"ccache gcc -m32" ./configure --x-libraries=/usr/X11R6/lib is creating a conflict so maybe the right configuration for a x86_64 linux with cache may be ( without having tested) CC"ccache gcc" ./configure --x-libraries=/usr/X11R6/lib.
Adding some notes on the respective page may be helpful
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #18 from Austin English austinenglish@gmail.com 2007-12-30 19:29:21 --- It's a wiki, if you see something wrong in it, feel free to edit it. But please do not do so unless you've tested the workaround and know it works.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #19 from Yolande Haneder yolande@haneder.biz 2007-12-31 13:15:34 --- The workaround is not working. Any option with ccache is crashing the compiling.
I fear that even with this problem fixed, the road to fixing the bug itself will not be over (it seemed to be already the 3rd bug for the same error). I tried reverting the problematic patch with wine 0.9.52. There is not setup error at "publishing informations", there is an exception at "registering components", further down the road. I however start to get menu icons for the different parts of the software.
I will try in a couple of days with a new GIT and with this patch reverted. A comment on another bug made me think of something that could help me. In the change, there is an error if the user name is not the same as the one of the application. Java is probably working as root and I am using Wine as an user. Is there not an error of Java being prohibited to respond because Java does not have the same user as Trados in wine? Is it not possible to make an exception for root in this patch?
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #20 from Dan Kegel dank@kegel.com 2008-01-01 00:19:07 --- How could Java be running as root?
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #21 from Yolande Haneder yolande@haneder.biz 2008-01-01 01:23:09 --- Java for Linux is being installed with the Suse package of Novell even before the user is being created. Thus it does not seem logic that it is named as the user.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #22 from Yolande Haneder yolande@haneder.biz 2008-01-01 03:03:45 --- Dear Dan,
Could you also test removing this commit for your test on bug 10984? I have the feeling it is also affecting Java for Windows, since I am making a new regression test with the first commit reverted to get the commit for error 2 and I just noticed that with this commit reverted, Java is still showing the error as in a case of a bad bisect of bug 10984, however the installation is proceeding through. I am hoping to have the other regression test by the end of the day.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #23 from Yolande Haneder yolande@haneder.biz 2008-01-01 06:54:51 --- Update:
Doing a regression test on the new error it is showing me the same commit (the only one remaining since I had reverted this commit on my git). Thus I may conclude that the new error is not a regression but an error that was previously hidden and now occurring.
The new error will be then subject of a new bug once this regression is fixed and I can officially send logs about it.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #24 from Dan Kegel dank@kegel.com 2008-01-01 14:19:37 --- Just because Java for Linux was installed as root doesn't mean it runs as root when you use it.
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #25 from Yolande Haneder yolande@haneder.biz 2008-01-02 12:19:14 --- I add you two pictures: one of the characteristics of Java and one of Trados. "Dateibesitzer" means file owner. You will see that one is root and the other is Yolande.
On the patch you can find
if (st.st_uid != getuid()) fatal_error( "%s is not owned by you\n", config_dir ); +#endif
Is there not a problem about it?
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #26 from Yolande Haneder yolande@haneder.biz 2008-01-02 12:20:59 --- Created an attachment (id=9986) --> (http://bugs.winehq.org/attachment.cgi?id=9986) file ownership of Trados
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #27 from Yolande Haneder yolande@haneder.biz 2008-01-02 12:23:29 --- Created an attachment (id=9987) --> (http://bugs.winehq.org/attachment.cgi?id=9987) File ownership of Java
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #28 from Austin English austinenglish@gmail.com 2008-06-04 11:29:38 --- Is this still an issue in 1.0-rc3 or newer wine?
http://bugs.winehq.org/show_bug.cgi?id=10584
--- Comment #29 from Yolande Haneder yolande@haneder.biz 2008-06-04 11:34:49 --- Yes and no. I would say no issue because the exception is gone. I would say yes because it's still hanging at the same place depending on the wine versions because assemblies are still not registering correctly.
You can close this bug - the new error message is very different.
http://bugs.winehq.org/show_bug.cgi?id=10584
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution| |FIXED
--- Comment #30 from Austin English austinenglish@gmail.com 2008-06-04 12:12:47 --- Fixed then. Please open a new bug for that problem.
http://bugs.winehq.org/show_bug.cgi?id=10584
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #31 from Alexandre Julliard julliard@winehq.org 2008-06-06 10:40:13 --- Closing bugs fixed in 1.0-rc4.
http://bugs.winehq.org/show_bug.cgi?id=10584
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Version|CVS/GIT |unspecified
http://bugs.winehq.org/show_bug.cgi?id=10584
Dan Kegel dank@kegel.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Platform|HP |PC
--- Comment #32 from Dan Kegel dank@kegel.com 2009-05-29 19:00:19 --- Bugs marked "HP, Linux" probably should have been marked "PC, Linux" as HP really is for HP's HP-PA or Itanium systems, and very few people have those.
-
wine-bugs＠winehq.org