Hi, just tested the new ParaView 3.6.1 ebuild on my amd64 machine, and the compile process fails with ================== *snip* ================== Linking CXX shared library ../../../../bin/libvtkQtChart.so [ 5%] Built target vtkQtChart make: *** [all] Error 2 /usr/libexec/paludis/utils/emake: emake returned error 2 ================== *snap* ================== What can I do to debug? Cheers, Nico Reproducible: Always
Ah, and compiling with VERBOSE=1 gives ================== *snip* ================== cd /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/build/VTK/GUISupport/Qt/Chart && /usr/bin/cmake -E cmake_symlink_library ../../../../bin/libvtkQtChart.so.pv3.6 ../../../../bin/libvtkQtChart.so.pv3.6 ../../../../bin/libvtkQtChart.so make[2]: Leaving directory `/var/tmp/paludis/sci-visualization-paraview-3.6.1/work/build' /usr/bin/cmake -E cmake_progress_report /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/build/CMakeFiles 43 44 [ 5%] Built target vtkQtChart make[1]: Leaving directory `/var/tmp/paludis/sci-visualization-paraview-3.6.1/work/build' make: *** [all] Error 2 ================== *snap* ==================
Please post your emerge --info and the complete build log obtained via VERBOSE=TRUE MAKEOPTS="-j1" emerge paraview As a matter of fact, this may be related to a parallel build failure and simply adding MAKEOPTS="-j1" may already help. Best, Markus
Hm, when I do "-j1", the build fails even earlier with =================== *snip* =================== /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/VTK/Common/vtkConditionVariable.cxx:27: error: conflicting declaration 'typedef int pthread_condattr_t' /usr/include/bits/pthreadtypes.h:136: error: 'pthread_condattr_t' has a previous declaration as 'typedef union pthread_condattr_t pthread_condattr_t' make[2]: *** [VTK/Common/CMakeFiles/vtkCommon.dir/vtkConditionVariable.o] Error 1 =================== *snap* =================== May that be b/c I'm running Qt 4.5.1?
Please post your emerge --info! This looks like a pthread issue and is not Qt related. Best, Markus
Can you also please try with USE="threads". It might be that I have to force pthreads in the ebuild and remove this useflag altogether. Thanks, Markus
I had the same problem. After setting VERBOSE=TRUE MAKEOPTS="-j1" it fails a little bit later with: [ 45%] Converting Tcl test cd /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/VTK/Filtering/Testing/Python && /usr/bin/python /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/VTK/Utilities/vtkTclTest2Py/vtkTclToPyConvertor.py -l /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/VTK/Filtering/Testing/Python/vtkTestsToConvert -t /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/VTK/Filtering/Testing/Python/conversion_complete -A /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/VTK/Utilities/vtkTclTest2Py -A /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/VTK/Wrapping/Python -A /var/tmp/portage/sci-visualization/paraview-3.6.1/work/build/bin/. Converting /var/tmp/portage/sci-visualization/paraview-3.6.1/work/ParaView3/VTK/Filtering/Testing/Tcl/closedSplines.tcl Conversion failed! make[2]: *** [VTK/Filtering/Testing/Python/conversion_complete] Error 1 It happens at three systems x86_64 and i686.
What is the version of Tcl/Tk?
Created attachment 199153 [details, diff] disable faulty python tcl tests Please try the attached patch and see if the issue goes away. Thanks, Markus
Okay, I got around my original compiling error by enablig the 'threads' flag. The build fails later, however, at ==================== *snip* ==================== [ 48%] Building CXX object Utilities/Xdmf2/libsrc/CMakeFiles/Xdmf.dir/XdmfH5Driver.o /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/Utilities/Xdmf2/libsrc/XdmfH5Driver.cxx:182: error: braces around scalar initializer for type 'H5FD_mem_t' /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/Utilities/Xdmf2/libsrc/XdmfH5Driver.cxx:182: error: invalid conversion from 'haddr_t (*)(const H5FD_t*, H5FD_mem_t)' to 'herr_t (*)(H5FD_t*, haddr_t)' /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/Utilities/Xdmf2/libsrc/XdmfH5Driver.cxx:182: error: invalid conversion from 'herr_t (*)(H5FD_t*, H5FD_mem_t, haddr_t)' to 'haddr_t (*)(H5FD_t*)' /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/Utilities/Xdmf2/libsrc/XdmfH5Driver.cxx:182: error: invalid conversion from 'haddr_t (*)(const H5FD_t*)' to 'herr_t (*)(H5FD_t*, hid_t, void**)' /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/Utilities/Xdmf2/libsrc/XdmfH5Driver.cxx:182: error: invalid conversion from 'herr_t (*)(H5FD_t*, H5FD_mem_t, hid_t, haddr_t, size_t, void*)' to 'herr_t (*)(H5FD_t*, H5FD_mem_t, hid_t, haddr_t, size_t, const void*)' /var/tmp/paludis/sci-visualization-paraview-3.6.1/work/ParaView3/Utilities/Xdmf2/libsrc/XdmfH5Driver.cxx:182: error: invalid conversion from 'herr_t (*)(H5FD_t*, H5FD_mem_t, hid_t, haddr_t, size_t, const void*)' to 'herr_t (*)(H5FD_t*, hid_t, unsigned int)' make[2]: *** [Utilities/Xdmf2/libsrc/CMakeFiles/Xdmf.dir/XdmfH5Driver.o] Error 1 make[1]: *** [Utilities/Xdmf2/libsrc/CMakeFiles/Xdmf.dir/all] Error 2 make: *** [all] Error 2 /usr/libexec/paludis/utils/emake: emake returned error 2 ==================== *snap* ====================
Please re-sync portage and try again. The threads issue plus the hdf5 issue in your last post should already be fixed in the most recent ebuild. If you run into the error reported in comment #6 please try the patch. If compilation still fails please post the latest error. BTW: You emerge --info would be extremely helpful. Any bug report should have it and it is VERY IMPORTANT for me to figure out what is going on. Thanks, Markus
I've had the same error as described in comment #0, however compiling paraview with USE="threads" worked flawlessly (I didn't set HDF5)
For everybody having the pthread (USE="thread") or hdf5 issue please re-sync portage. These issues have been fixed already in the most recent ebuild. Best, Markus
Created attachment 199172 [details, diff] updated patch Please use this updated patch instead.
Okay, with the latest ebuild and "-j1" it compiles and installs fine for me, so I guess we could in principle mark this one fixed.
Thanks for the feedback, Nico! I am glad it compiles for you now. I'll close the bug once the others confirm that they can compile it now also. Best, Markus
Since I haven't heard any any additional failure reports I presume we're good and I'll close this one. cheers, Markus