Qt-interest Archive, May 2008
Problems compiling Qt 4.4.0 on linux
Message 1 in thread
Hi everybody,
I've been trying to build Qt 4.4.0 on a handful of linux platforms
without much success. So far I've had no problems compiling on
Windows, Mac OS X, and RHEL 5 (x86/x64). The most common error
(although one among many) is that I get an internal compiler error
when compiling
JavaScriptCore/pcre_ucp_findchar.c
on RHEL4 x86 using gcc 3.4.3. I've tried using gcc 4.1.2 on that
machine as well but get a different error when linking libQtCore:
/opt/local/lib/gcc/i686-pc-linux-gnu/4.1.2/../../../libstdc+
+.a(eh_globals.o): In function `__gnu_internal::get_global()':
../../../../gcc-4.1.2/libstdc++-v3/libsupc++/eh_globals.cc:58:
undefined reference to `___tls_get_addr'
I used the default Qt configuration on this machine as well as
compiling with -no-phonon to see if that would make a difference. I've
run into these same problems on RHEL3 x86/x64, RHEL4 x84/x64 and suse
9.1 x86/x64. My inclination is to think that gcc 3.4.3 just won't work
with Qt 4.4.0, but I was hoping that maybe somebody out there had seen
and resolved this before.
Thanks,
Joe
Joseph Corkery, M.D.
Principal Developer
OpenEye Scientific Software, Inc.
222 Third Street, Suite 3120
Cambridge, MA 02142
T: 505-473-7385 x76
F: 617-374-6575
--
[ signature omitted ]
Message 2 in thread
Hi,
> I've been trying to build Qt 4.4.0 on a handful of linux platforms
> without much success. So far I've had no problems compiling on Windows,
> Mac OS X, and RHEL 5 (x86/x64). The most common error (although one
> among many) is that I get an internal compiler error when compiling
>
> JavaScriptCore/pcre_ucp_findchar.c
>
> on RHEL4 x86 using gcc 3.4.3. I've tried using gcc 4.1.2 on that machine
A compiler internal error is a compiler bug, so it should be fixed by the
compiler vendor. Qt might sometimes provide a workaround if the problem
happens on a very common platform and no bug fixes are made available for the
compiler. Note that before anything is changed in Qt, a bug report must have
been filed so that the problem is known to Trolltech:
http://trolltech.com/bugreport-form
The latest compiler available for RHEL 4 seems to be:
gcc-3.4.6-9
Please upgrade the compiler to the latest available version on your platform.
The above bug might have been fixed.
Note that the above error happens when compiling the WebKit module. You may
disable WebKit if you don't need it:
configure -no-webkit
> as well but get a different error when linking libQtCore:
>
> /opt/local/lib/gcc/i686-pc-linux-gnu/4.1.2/../../../libstdc++.a(eh_globals.o):
> In function `__gnu_internal::get_global()':
> ../../../../gcc-4.1.2/libstdc++-v3/libsupc++/eh_globals.cc:58: undefined
> reference to `___tls_get_addr'
This really is an issue with teh compiler. I noticed that libstdc++ is a
static library, not a shared library. This doesn't look like a default gcc
installation.
--
[ signature omitted ]
Message 3 in thread
Hi Joe,
I don't know if it will help, but I recently built Qt4.4 on Kubuntu 8.04
using gcc 3.4.6. However, I had an error message saying that
"-fvisibility=hidden" was an unrecognized option for the compiler. I
solved this erasing this option from all the makefiles.
Good luck,
Susana
Joe Corkery escribió:
> Hi everybody,
>
> I've been trying to build Qt 4.4.0 on a handful of linux platforms
> without much success. So far I've had no problems compiling on
> Windows, Mac OS X, and RHEL 5 (x86/x64). The most common error
> (although one among many) is that I get an internal compiler error
> when compiling
>
> JavaScriptCore/pcre_ucp_findchar.c
>
> on RHEL4 x86 using gcc 3.4.3. I've tried using gcc 4.1.2 on that
> machine as well but get a different error when linking libQtCore:
>
> /opt/local/lib/gcc/i686-pc-linux-gnu/4.1.2/../../../libstdc++.a(eh_globals.o):
> In function `__gnu_internal::get_global()':
> ../../../../gcc-4.1.2/libstdc++-v3/libsupc++/eh_globals.cc:58:
> undefined reference to `___tls_get_addr'
>
> I used the default Qt configuration on this machine as well as
> compiling with -no-phonon to see if that would make a difference. I've
> run into these same problems on RHEL3 x86/x64, RHEL4 x84/x64 and suse
> 9.1 x86/x64. My inclination is to think that gcc 3.4.3 just won't work
> with Qt 4.4.0, but I was hoping that maybe somebody out there had seen
> and resolved this before.
>
> Thanks,
> Joe
>
> Joseph Corkery, M.D.
> Principal Developer
> OpenEye Scientific Software, Inc.
> 222 Third Street, Suite 3120
> Cambridge, MA 02142
> T: 505-473-7385 x76
> F: 617-374-6575
>
>
>
> --
> To unsubscribe - send a mail to qt-interest-request@xxxxxxxxxxxxx with
> "unsubscribe" in the subject or the body.
> List archive and information: http://lists.trolltech.com/qt-interest/
>
>
> __________ Información de NOD32, revisión 3128 (20080523) __________
>
> Este mensaje ha sido analizado con NOD32 antivirus system
> http://www.nod32.com
>
>
>
--
[ signature omitted ]
Message 4 in thread
Hi,
> I don't know if it will help, but I recently built Qt4.4 on Kubuntu 8.04
> using gcc 3.4.6. However, I had an error message saying that
> "-fvisibility=hidden" was an unrecognized option for the compiler. I
> solved this erasing this option from all the makefiles.
This seems unrelated to the original poster's problem.
It's interesting that you had this "fvisibility" issue on your platform
though. In theory "configure" auto-detects whether "fvisibility" is supported
or not and it shouldn't be supported with GCC 3 indeed:
$ cat config.tests/unix/fvisibility.test
[...]
cat >>fvisibility.c << EOF
__attribute__((visibility("default"))) void blah();
#if !defined(__GNUC__)
# error "Visiblility support requires GCC"
#elif __GNUC__ < 4
# error "GCC3 with backported visibility patch is known to miscompile Qt"
#endif
EOF
[...]
$
Auto-detection seems not to have worked in your case. What is the output of
the following commands?
config.tests/unix/fvisibility.test g++ yes
By the way, instead of erasing the option from all the makefiles, I think you
could have specified:
configure -no-reduce-relocations
--
[ signature omitted ]
Message 5 in thread
Hi,
Dimitri escribió:
> Hi,
>
>> I don't know if it will help, but I recently built Qt4.4 on Kubuntu
>> 8.04 using gcc 3.4.6. However, I had an error message saying that
>> "-fvisibility=hidden" was an unrecognized option for the compiler. I
>> solved this erasing this option from all the makefiles.
>
> This seems unrelated to the original poster's problem.
>
> It's interesting that you had this "fvisibility" issue on your
> platform though. In theory "configure" auto-detects whether
> "fvisibility" is supported or not and it shouldn't be supported with
> GCC 3 indeed:
> $ cat config.tests/unix/fvisibility.test
> [...]
> cat >>fvisibility.c << EOF
> __attribute__((visibility("default"))) void blah();
> #if !defined(__GNUC__)
> # error "Visiblility support requires GCC"
> #elif __GNUC__ < 4
> # error "GCC3 with backported visibility patch is known to miscompile Qt"
> #endif
> EOF
> [...]
> $
>
> Auto-detection seems not to have worked in your case. What is the
> output of the following commands?
> config.tests/unix/fvisibility.test g++ yes
>
I have run the test, and here is the result:
$ config.tests/unix/fvisibility.test g++ yes
Symbol visibility control enabled.
I think that maybe it is because I have more than one version of gcc in
my system. At first the version installed was 4.2, but I had to build
sources that only accepted gcc version 3, so I installed it and set it
as the default with update-alternatives.
> By the way, instead of erasing the option from all the makefiles, I
> think you could have specified:
> configure -no-reduce-relocations
>
I didn't see it in the help of configure... :S
Thank you!
Susana
--
[ signature omitted ]
Message 6 in thread
Susana escribió:
> Hi,
>
> Dimitri escribió:
>> Hi,
>>
>>> I don't know if it will help, but I recently built Qt4.4 on Kubuntu
>>> 8.04 using gcc 3.4.6. However, I had an error message saying that
>>> "-fvisibility=hidden" was an unrecognized option for the compiler. I
>>> solved this erasing this option from all the makefiles.
>>
>> This seems unrelated to the original poster's problem.
>>
>> It's interesting that you had this "fvisibility" issue on your
>> platform though. In theory "configure" auto-detects whether
>> "fvisibility" is supported or not and it shouldn't be supported with
>> GCC 3 indeed:
>> $ cat config.tests/unix/fvisibility.test
>> [...]
>> cat >>fvisibility.c << EOF
>> __attribute__((visibility("default"))) void blah();
>> #if !defined(__GNUC__)
>> # error "Visiblility support requires GCC"
>> #elif __GNUC__ < 4
>> # error "GCC3 with backported visibility patch is known to miscompile
>> Qt"
>> #endif
>> EOF
>> [...]
>> $
>>
>> Auto-detection seems not to have worked in your case. What is the
>> output of the following commands?
>> config.tests/unix/fvisibility.test g++ yes
>>
> I have run the test, and here is the result:
>
> $ config.tests/unix/fvisibility.test g++ yes
> Symbol visibility control enabled.
>
> I think that maybe it is because I have more than one version of gcc
> in my system. At first the version installed was 4.2, but I had to
> build sources that only accepted gcc version 3, so I installed it and
> set it as the default with update-alternatives.
>> By the way, instead of erasing the option from all the makefiles, I
>> think you could have specified:
>> configure -no-reduce-relocations
>>
> I didn't see it in the help of configure... :S
>
> Thank you!
>
> Susana
>
> --
> To unsubscribe - send a mail to qt-interest-request@xxxxxxxxxxxxx with
> "unsubscribe" in the subject or the body.
> List archive and information: http://lists.trolltech.com/qt-interest/
>
>
Hi again,
After uninstalling the other versions of gcc, the test has changed the
outcome:
$ config.tests/unix/fvisibility.test g++ yes
Symbol visibility control disabled.
Regards,
Susana
--
[ signature omitted ]
Message 7 in thread
Hello Everybody,
I want to compile Qt-4.4.0 with Phonon support on a Windows 2000 machine
with MSVC 2005. I figured out that
DirectX 9.0c is not supported on Win2k. As estimated my Qt build broke
when compiling the phonon lib because some DirectX dependencies were not
found.
Therefore I downloaded and installed DirectX 9.0 SDK December 2004 as I
think this is the latest version that comes with Win2K Support. I also
have the Platform SDK Win 2003 Server R2 Release installed on that
machine. All include, lib and path environment variables seem to be
correct. But now the Qt build breaks with the following message:
cl -c -FIqt_gui_pch.h -Yuqt_gui_pch.h
-Fptmp\obj\debug_shared\QtGuid_pch.pch -nologo -Zm200 -Zc:wchar_t- -Zi
-MDd -W3 -w34100 -w34189 -GR -EHsc -DQT_SHA RED -DQT_THREAD_SUPPORT
-DUNICODE -DWIN32 -DQT_LARGEFILE_SUPPORT -DQT_BUILD_GUI_LIB
-DQT_NO_USING_NAMESPACE -DQT_MAKEDLL -DQT_NO_CAST_TO_ASCII
-DQT_ASCII_CAST_W ARNINGS -DQT3_SUPPORT -DQT_MOC_COMPAT
-D_USE_MATH_DEFINES -DPNG_NO_ASSEMBLER_CODE -DQT_RASTER_IMAGEENGINE
-DQT_RASTER_PAINTENGINE -DQT_NO_CUPS -DQT_NO_LPR -DQT_ HAVE_MMX
-DQT_HAVE_3DNOW -DQT_HAVE_SSE -DQT_HAVE_MMXEXT -DQT_HAVE_SSE2
-DQT_NO_FREETYPE -DQT_NO_OPENTYPE -DQT_NO_STYLE_MAC
-DQT_NO_STYLE_WINDOWSCE -DQT_NO_STYLE _WINDOWSMOBILE
-DQ_INTERNAL_QAPP_SRC -DQT_DLL -DQT_CORE_LIB -I"..\..\include\QtCore"
-I"..\..\include\QtCore" -I"..\..\include" -I"..\..\include\QtGui"
-I"tmp\r cc\debug_shared" -I"tmp" -I"..\3rdparty\wintab"
-I"..\3rdparty\libpng" -I"..\3rdparty\zlib" -I"..\3rdparty\harfbuzz\src"
-I"dialogs" -I"s:\swamp\RBTools\SWAMP\l
ibs\qt-win-opensource-src-4.4.0\include\ActiveQt"
-I"tmp\moc\debug_shared" -I"." -I"..\..\mkspecs\win32-msvc2005"
-Fotmp\obj\debug_shared\ @C:\Temp\nmBD.tmp qpaintengine_d3d.cpp
painting\qpaintengine_d3d.cpp(3438) : error C2065:
'D3DXFX_NOT_CLONEABLE' : undeclared identifier
painting\qpaintengine_d3d.cpp(3438) : error C2065:
'D3DXSHADER_OPTIMIZATION_LEVEL3' : undeclared identifier NMAKE : fatal
error U1077: '"C:\Program Files\Microsoft Visual Studio
8\VC\BIN\cl.EXE"' : return code '0x2'
Stop.
I think this is an issue with Direct3D?
Does anyone can help me? How can I compile Qt 4.4 on Win2k? Under
Windows XP I do not have any problems with the Libraries.
Greetings,
Thomas Berger
*******************************************
Harman Becker Automotive Systems GmbH
Geschaeftsfuehrung: Dr. Wolfgang Ptacek - Michael Mauser - Regis Baudot
Sitz der Gesellschaft: Karlsbad - Registergericht: Mannheim HRB 361395
*******************************************
Diese E-Mail enthaelt vertrauliche und/oder rechtlich geschuetzte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtuemlich erhalten haben, informieren Sie bitte sofort den Absender und loeschen Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail ist nicht gestattet.
This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and delete this e-mail. Any unauthorized copying, disclosure or distribution of the contents in this e-mail is strictly forbidden.
*******************************************
--
[ signature omitted ]
Message 8 in thread
Actually, I ran into this problem as well in one of my many iterations
of turning options off. I think maybe when I built with either webkit or
phonon off I got the same error.
Joe
Dimitri wrote:
> Hi,
>
>> I don't know if it will help, but I recently built Qt4.4 on Kubuntu
>> 8.04 using gcc 3.4.6. However, I had an error message saying that
>> "-fvisibility=hidden" was an unrecognized option for the compiler. I
>> solved this erasing this option from all the makefiles.
>
> This seems unrelated to the original poster's problem.
>
> It's interesting that you had this "fvisibility" issue on your
> platform though. In theory "configure" auto-detects whether
> "fvisibility" is supported or not and it shouldn't be supported with
> GCC 3 indeed:
> $ cat config.tests/unix/fvisibility.test
> [...]
> cat >>fvisibility.c << EOF
> __attribute__((visibility("default"))) void blah();
> #if !defined(__GNUC__)
> # error "Visiblility support requires GCC"
> #elif __GNUC__ < 4
> # error "GCC3 with backported visibility patch is known to miscompile Qt"
> #endif
> EOF
> [...]
> $
>
> Auto-detection seems not to have worked in your case. What is the
> output of the following commands?
> config.tests/unix/fvisibility.test g++ yes
>
> By the way, instead of erasing the option from all the makefiles, I
> think you could have specified:
> configure -no-reduce-relocations
>
--
[ signature omitted ]
Message 9 in thread
I had the same problem with RHEL 3 and gcc 3.4.x After installing gcc 4+
everything went fine.
--
[ signature omitted ]
