Package Details: paraview 5.4.1-1

Git Clone URL: (read-only)
Package Base: paraview
Description: Parallel Visualization Application using VTK
Upstream URL:
Licenses: custom
Submitter: stephane
Maintainer: deezy (eolianoe)
Last Packager: eolianoe
Votes: 66
Popularity: 1.343649
First Submitted: 2012-12-11 01:37
Last Updated: 2017-08-22 19:43

Latest Comments

ArchangeGabriel commented on 2017-08-13 09:13

Hum yes, that the kind of issue you get with AUR packages when rebuild land in Arch.

I’m still planning to move this (as well as cgns and ospray) in [community], but I’m still unsure of the way to go w.r.t. ISA. ospray requires at least SSE4, and an SSE4-enabled embree, but Arch is still SSE2 (a bit off topic, but something we might want to look at when i686 will be gone)… I might end up building a embree-sse4 package, but people should be advised to install embree-isa and ospray-isa to get the most of them.

lahwaacz commented on 2017-08-11 14:13

I had a couple of warnings like

/usr/bin/ld: warning:, needed by /home/klinkovsky/build/builddir/paraview/src/build/lib/, not found (try using -rpath or -rpath-link)

in the rebuild, but they went away when I removed the build directory and compiled cgns as well as paraview from scratch.

lonaowna commented on 2017-08-11 13:18

Anyone else having trouble with hdf5 1.10.1? I tried rebuilding paraview, but that didn't change anything.

/usr/lib/paraview-5.4/paraview: error while loading shared libraries: cannot open shared object file: No such file or directory

eolianoe commented on 2017-08-01 17:35

@tomracing: maybe you should rebuild ospray to fix the undefined references.

tomracing commented on 2017-07-29 17:11

Package failed to build with the following errors after a few notifications of undefined items from libospray refering to some functions.
collect2: error: ld returned 1 exit status
[4683/18298] Building CXX object VTK/Render...ringParallelLIC.dir/vtkPPixelTransfer.cxx.o
ninja: build stopped: subcommand failed.
==> ERROR: A failure occurred in build().
:: failed to build paraview package(s)

Any idea if it is the pkgbuild or the dependencies going wrong here?

lahwaacz commented on 2017-07-13 10:26

Figured it out - had to use the "Cell Data to Point Data" filter first...

lahwaacz commented on 2017-07-12 13:22

When I open any dataset in paraview and click on the "Apply" button, the contour filter is still greyed out and unusable. What should I do to make it available? All tutorials start with something like "click on Filters/Common/Contour", but don't consider that it is not available.

xantares commented on 2017-07-08 07:37

here's what I had to do to fix this:

move paraview module into site-packages:
sudo mv /usr/lib/site-packages/paraview/ /usr/lib/python3.6/site-packages/

remove the vtk dir without binaries
sudo rm -r /usr/lib/python3.6/site-packages/paraview/vtk

move vtk files with binaries into paraview module:
sudo mv /usr/lib/paraview-5.4/site-packages/vtk /usr/lib/python3.6/site-packages/paraview

xantares commented on 2017-07-08 07:33

exactly, if the python module was installed correctly you wouldnt have to set PYTHONPATH, like any other archlinux python package.

lahwaacz commented on 2017-07-08 07:27

@xantares: There is nothing in /usr/lib/site-packages on my system. To let python see /usr/lib/paraview-5.4/site-packages/ you need to set PYTHONPATH accordingly:

export PYTHONPATH=/usr/lib/paraview-5.4/site-packages:/usr/lib/paraview-5.4/site-packages/vtk

xantares commented on 2017-07-08 07:15


the python module cannot be imported:
In [2]: import paraview.annotation

theres python stuff in

but some binaries in /usr/lib/paraview-5.4/site-packages/

but I figure these should be moved into /usr/lib/python3.6/site-packages

setting VTK_INSTALL_LIBRARY_DIR=lib will help

GloW commented on 2017-06-23 14:58

@xantares : most ParaView use it as Gui, that is true, and that's why the main paraview package should be aimed at the main userbase in my opinion. It is @eolianoe package, so that's his decision anyway.

xantares commented on 2017-06-23 13:10

@GloW why rebuilding paraview if it's available in the packages? plus it takes forever to build. I guess you guys at paraview are not used to the fact that we use it as a lib rather than a gui app.

GloW commented on 2017-06-23 11:29

@eolianoe : Sure you can (considering VTK_NO_PYTHON_THREAD is OFF), it will not disable any feature, but it does have an overhead for each python call.

eolianoe commented on 2017-06-23 11:23

@GloW: thanks for the explanation, if it does not disable any feature, it can be easily enable in the PKGUILD.
Yeah, an expert will build its own version but I'm trying to have most of the option enabled.

GloW commented on 2017-06-23 11:05

@eolianoe @xantares :
-DVTK_PYTHON_FULL_THREADSAFE=ON is a very specific option meant to allow usage of external python interpreter within a ParaView based application.

Anyone willing to build their own ParaView based application would build ParaView themselves anyway, and not use the aur package. So no reason to activate it imo.

source : Am the developper of this option :) .

eolianoe commented on 2017-06-23 09:09

@xantares: could you explain the benefits of the full_threadsafe option? For the install directory, I'm more in favour of following the default set by upstream.

xantares commented on 2017-06-22 19:03

could you add:

so to avoid to have to set LD_LIBRARY_PATH


antran commented on 2017-05-31 02:10

@eolianoe: Yes. I rebuilt it and now it works fine. Thank you very much for your support

Hoang An

eolianoe commented on 2017-05-30 12:12

@antran: just rebuild it and paraview will find the right libraries

antran commented on 2017-05-30 03:11

Hello everyone,

I installed the latest version of paraview on my laptop but I cannot open it. I got the following errors on Konsole

/usr/lib/paraview-5.3/paraview: error while loading shared libraries: cannot open shared object file: No such file or directory

Can anyone help me? Thank you in advance.

Hoang An

eolianoe commented on 2017-04-24 14:57

@Azrael: the openssl-1.0 package is already resolved by the qt5 dependencies

Azrael commented on 2017-04-24 12:39


for the build to works I had to add
* openssl-1.0
as a dependency

The current version is now openssl-1.1.1 and so the build fail because xmlpattern can't find otherwise.


xantares commented on 2017-03-22 21:27

This package does not install headers, maybe switch the PARAVIEW_INSTALL_DEVELOPMENT_FILES cmake option ?

eolianoe commented on 2016-11-28 16:35

In case you want more features that are not enabled in this package (for instance due to library conflicts), you can look at the binary package [a] or at the new superbuild package [b].


GloW commented on 2016-11-27 21:17

yes, no data are loaded until you hit apply, paraview need data to give you available filters.

@juliohm, in case the package is unstable, you can use the binaries or compile yourself, both are quite stable.

lahwaacz commented on 2016-11-27 20:57

I've only used Paraview from this package, so I don't really know.

juliohm commented on 2016-11-27 15:21

Indeed @Iahwaacz, thank you for sharing. Is this expected behavior?

lahwaacz commented on 2016-11-26 22:08

All filters ungray themselves when I click on the "Apply" button in the "Properties" tab after loading a file.

juliohm commented on 2016-11-26 21:36

All the ParaView filters are disabled (grayish) in the current build. Could you please fix this? I am always afraid of updating this package, every time a new major issue.

juliohm commented on 2016-10-19 22:52

Thank you @eolianoe, will do.

eolianoe commented on 2016-10-19 20:37

@juliohm: This was fixed in 5.1.2-3, you can install the new version

juliohm commented on 2016-10-19 18:52

Sorry @deezy, I didn't receive a notification on my email when you commented. Fortunately, @eolianoe has answered your question. What is the current status of the issue? Can I reinstall ParaView?

eolianoe commented on 2016-10-14 07:42

@deep_thought: I can reproduce it in a clean chroot, so I think it's a missing dependency. I'll try to find which one it is.

deep_thought commented on 2016-10-14 04:36

compilation fails near the end with:
/bin/sh: line 0: cd: /tmp/paraview/src/build/Plugins/SurfaceLIC/doc: No such file or directory
make[2]: *** [Plugins/SurfaceLIC/CMakeFiles/SurfaceLIC.dir/build.make:79: Plugins/SurfaceLIC/doc/SurfaceLIC.qch] Error 1
make[1]: *** [CMakeFiles/Makefile2:47179: Plugins/SurfaceLIC/CMakeFiles/SurfaceLIC.dir/all] Error 2
Am I missing a dependency?
It's my minimal build environment: only base and base-devel are installed.

kikislater commented on 2016-10-13 16:39

Erreur de segmentation (core dumped)

eolianoe commented on 2016-10-11 07:45

@deezy: The conflict occurs with community/vtk.

deezy commented on 2016-10-11 07:40

@juliohm There is no conflict with hdf5 from Extra. Which package does that file belong to? (pacman -Qo)

juliohm commented on 2016-10-10 23:34

Can you please remove the file /usr/share/cmake/hdf5/libhdf5.settings from this build? It is already owned by another package in Arch and I cannot update the system because of the conflict.

GloW commented on 2016-10-06 08:15

@alen12345 : it is located in /usr/bin/qhelpgenerator

alen12345 commented on 2016-10-06 05:22

Dear mantainer, in a recent update, /usr/lib/qt/bin/qhelpgenerator has been removed due to a conflict with qt5; paraview is now failing to build.

NOhs commented on 2016-09-17 22:51

@maintainer: Paraview release notes say: "ParaView now comes with a built in ray traced rendering capability." Which is a reference to the new OSPray feature. Will this be added in a future update of this package? As OSPray comes with the default compiled versions I think the package in the AUR called "paraview" should probably stick to the default and include it. Would be awesome :).

ugis commented on 2016-08-15 11:17

Hi everyone,

I have successfully installed the paraview on my laptop, however, I have an issue with openGL rendering. Everything works overall, but surface of a given body is randomly transparent - i.e., parts of the surface seem transparent, such that part of the body can be seen from inside.

Could this be bug of the package build? I have checked openGL rendering using glxgears and unigine-heaven, and I don't see any problems there.


melanopsis commented on 2016-06-20 23:00

This won't compile for me without the qt5-x11extras package

deezy commented on 2016-06-02 14:57

hdf5 (1.8.x) is included in the paraview source and is used as long hdf5-1.10.x (in Extra) and paraview/vtk are not really compatible.
With the next release, the new Extra package hdf5_18 may be used. But I have to test this first.
So at the moment there is no missing dependency.

hiaselhans commented on 2016-06-02 13:17

missing dependency: hdf5

lahwaacz commented on 2016-05-09 17:10

I'm not getting the 8th line:

deezy commented on 2016-05-09 16:03

@lahwaacz @juliohm
Added the Exec key to the desktop file. Should work now.

lahwaacz commented on 2016-05-09 15:34

See the specification:

"The Exec key is required if DBusActivatable is not set to true."

So the paraview.desktop does not comply to the standard.

juliohm commented on 2016-05-09 14:59

Everything is perfect, except the desktop file created under /usr/share/applications/paraview.desktop. When I try to open paraview from the Gnome menu it is not found with the message "Desktop file didn't specify Exec field". What is the exact difference between Exec and TryExec? Why to opt for the latter?

wilmbas commented on 2016-05-07 09:43

Build fails with recent version 6 of GCC. I am newbie in pushing updated .SRCINFO and modified PKGBUILD's. But patch and sha1sum is created and added to it. Since I failed to push it please find the patch concatenated to this comment. (Luckily, since I missed to have correct path strings towards VTK/CMake in my patch...newbie behaviour, sorry!)

Have fun, nice weekend. Regards wilmbas

--- VTK/CMake/GenerateExportHeader.cmake 2016-05-07 10:22:52.463297686 +0200
+++ VTK/CMake/ 2016-05-07 10:20:26.030649703 +0200
@@ -166,7 +166,7 @@
execute_process(COMMAND ${CMAKE_C_COMPILER} ARGS --version
OUTPUT_VARIABLE _gcc_version_info
ERROR_VARIABLE _gcc_version_info)
- string(REGEX MATCH "[345]\\.[0-9]\\.[0-9]*"
+ string(REGEX MATCH "[3-9]\\.[0-9]\\.[0-9]*"
_gcc_version "${_gcc_version_info}")
# gcc on mac just reports: "gcc (GCC) 3.3 20030304 ..." without the
# patch level, handle this here:
--- VTK/CMake/vtkCompilerExtras.cmake 2016-05-07 10:23:05.523533858 +0200
+++ VTK/CMake/ 2016-05-07 10:22:03.425744259 +0200
@@ -32,7 +32,7 @@
OUTPUT_VARIABLE _gcc_version_info
ERROR_VARIABLE _gcc_version_info)

- string (REGEX MATCH "[345]\\.[0-9]\\.[0-9]*"
+ string (REGEX MATCH "[3-9]\\.[0-9]\\.[0-9]*"
_gcc_version "${_gcc_version_info}")
if(NOT _gcc_version)
string (REGEX REPLACE ".*\\(GCC\\).*([34]\\.[0-9]).*" "\\1.0"

eolianoe commented on 2016-05-04 17:10

The paraview.desktop and paraview.png files can be deleted as the sources already provide those files and they are installed by `make install`.

andrea89g commented on 2016-05-03 12:22

same problem of chuckdaniels, i get these:
In file included from /tmp/yaourt-tmp-andrea/aur-paraview/src/build/VTK/ThirdParty/netcdf/vtknetcdf/config.h:7:0,
from /tmp/yaourt-tmp-andrea/aur-paraview/src/ParaView-v5.0.1-source/VTK/ThirdParty/netcdf/vtknetcdf/libsrc4/nc4internal.c:18:
/tmp/yaourt-tmp-andrea/aur-paraview/src/ParaView-v5.0.1-source/VTK/ThirdParty/netcdf/vtknetcdf/include/vtk_netcdf_mangle.h:229:30: error: conflicting types for ‘vtk_netcdf_nc4_rec_find_nc_type’
#define nc4_rec_find_nc_type vtk_netcdf_nc4_rec_find_nc_type
/tmp/yaourt-tmp-andrea/aur-paraview/src/ParaView-v5.0.1-source/VTK/ThirdParty/netcdf/vtknetcdf/libsrc4/nc4internal.c:383:1: note: in expansion of macro ‘nc4_rec_find_nc_type’
nc4_rec_find_nc_type(NC_GRP_INFO_T *start_grp, nc_type target_nc_typeid)
/tmp/yaourt-tmp-andrea/aur-paraview/src/ParaView-v5.0.1-source/VTK/ThirdParty/netcdf/vtknetcdf/include/vtk_netcdf_mangle.h:229:30: note: previous declaration of ‘vtk_netcdf_nc4_rec_find_nc_type’ was here
#define nc4_rec_find_nc_type vtk_netcdf_nc4_rec_find_nc_type
/tmp/yaourt-tmp-andrea/aur-paraview/src/ParaView-v5.0.1-source/VTK/ThirdParty/netcdf/vtknetcdf/libsrc4/nc4internal.h:370:17: note: in expansion of macro ‘nc4_rec_find_nc_type’
NC_TYPE_INFO_T *nc4_rec_find_nc_type(NC_GRP_INFO_T *start_grp, hid_t target_nc_typeid);
VTK/ThirdParty/netcdf/vtknetcdf/CMakeFiles/vtkNetCDF.dir/build.make:590: recipe for target 'VTK/ThirdParty/netcdf/vtknetcdf/CMakeFiles/vtkNetCDF.dir/libsrc4/nc4internal.c.o' failed
make[2]: *** [VTK/ThirdParty/netcdf/vtknetcdf/CMakeFiles/vtkNetCDF.dir/libsrc4/nc4internal.c.o] Error 1
CMakeFiles/Makefile2:8139: recipe for target 'VTK/ThirdParty/netcdf/vtknetcdf/CMakeFiles/vtkNetCDF.dir/all' failed
make[1]: *** [VTK/ThirdParty/netcdf/vtknetcdf/CMakeFiles/vtkNetCDF.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....

chuckdaniels commented on 2016-05-03 11:42

Seem that it is not starting with the last hdf5 package update. I cannot recompile the package either.

MartinDiehl commented on 2016-04-11 20:06

Using the hdf5-fortran-openmpi package, I get an error (see below) when opening XDMF files (which are opening an HDF5 files). I'm not really sure if this would need to be fixed by the paraview developers or in the Arch package, so please feel free to delete this hint if it's inappropriate.

Warning! ***HDF5 library version mismatched error***
The HDF5 header files used to compile this application do not match
the version used by the HDF5 library to which this application is linked.
Data corruption or segmentation faults may occur if the application continues.
This can happen when an application was compiled by one version of HDF5 but
linked with a different version of static or shared HDF5 library.
You should recompile the application or check your shared library related
settings such as 'LD_LIBRARY_PATH'.
'HDF5_DISABLE_VERSION_CHECK' environment variable is set to 1, application will
continue at your own risk.
Headers are 1.8.15, library is 1.8.16

eolianoe commented on 2016-03-08 08:39

Paraview can also be build against ffmpeg 3.0 using the patch from this bugreport [1]

If you want the export into pdf (or other vector format) remove the line `-DVTK_RENDERING_BACKEND:STRING=OpenGL2` (see mailing-list discussion [2])


heczis commented on 2016-03-07 14:07

On x86_64, I get:

Scanning dependencies of target vtkIOFFMPEG
[ 12%] Building CXX object VTK/IO/FFMPEG/CMakeFiles/vtkIOFFMPEG.dir/vtkFFMPEGWriter.cxx.o
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx: In member function ‘int vtkFFMPEGWriterInternal::Start()’:
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx:194:18: error: ‘PIX_FMT_YUVJ422P’ was not declared in this scope
c->pix_fmt = PIX_FMT_YUVJ422P;
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx:198:18: error: ‘PIX_FMT_BGR24’ was not declared in this scope
c->pix_fmt = PIX_FMT_BGR24;
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx:277:40: error: ‘avcodec_alloc_frame’ was not declared in this scope
this->rgbInput = avcodec_alloc_frame();
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx:283:36: error: ‘PIX_FMT_RGB24’ was not declared in this scope
int RGBsize = avpicture_get_size(PIX_FMT_RGB24, c->width, c->height);
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx: In member function ‘int vtkFFMPEGWriterInternal::Write(vtkImageData*)’:
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx:357:28: error: ‘PIX_FMT_RGB24’ was not declared in this scope
cc->width, cc->height, PIX_FMT_RGB24,
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx: In member function ‘int vtkFFMPEGWriterInternal::Start()’:
/home/jan/SOFT/paraview/src/ParaView-v5.0.0-source/VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx:326:53: warning: ignoring return value of ‘int avformat_write_header(AVFormatContext*, AVDictionary**)’, declared with attribute warn_unused_result [-Wunused-result]
avformat_write_header(this->avFormatContext, NULL);
VTK/IO/FFMPEG/CMakeFiles/vtkIOFFMPEG.dir/build.make:71: recipe for target 'VTK/IO/FFMPEG/CMakeFiles/vtkIOFFMPEG.dir/vtkFFMPEGWriter.cxx.o' failed
make[2]: *** [VTK/IO/FFMPEG/CMakeFiles/vtkIOFFMPEG.dir/vtkFFMPEGWriter.cxx.o] Error 1
CMakeFiles/Makefile2:7158: recipe for target 'VTK/IO/FFMPEG/CMakeFiles/vtkIOFFMPEG.dir/all' failed
make[1]: *** [VTK/IO/FFMPEG/CMakeFiles/vtkIOFFMPEG.dir/all] Error 2
Makefile:149: recipe for target 'all' failed
make: *** [all] Error 2
==> ERROR: A failure occurred in build().

capitalaslash commented on 2016-02-23 11:55

bug 14215 has finally been solved in vtk 7.0, the dependency should change to ffmpeg.
(actually, using ffmpeg-0.10 does not work, since headers are now stored in a different directory).

mbeaudoin commented on 2015-12-06 13:35

Would it be possible to enable the installation of the ParaView development files by adding the following compile option in PKGBUILD:


This would, for instance, facilitate the development of ParaView/Catalyst modules under Arch Linux.

Many thanks in advance.

mbeaudoin commented on 2015-12-06 13:27

@eleftg and @sachkris:

About the comment posted on 2015-03-05 06:19

I am getting the same compilation error on a 32-bit Arch Linux system.
A bug report was logged on the ParaView bug tracker

The fix is simple, just a simple type cast at line 765 for the file

We should have:
vector<Arm*> armpair(2,(Arm*)NULL);

instead of:
vector<Arm*> armpair(2,NULL);

Until this gets fixed in ParaView, a patch file would be necessary.

eolianoe commented on 2015-10-02 09:30

In case someone is interested I uploaded a Qt5 version of ParaView [1], feel free to test it.

sachkris commented on 2015-09-04 11:08

@eleftg and @alen12345
Sorry, I should have posted this earlier.

The compile error reported before was coming in the system where Arch linux was 32bit (i686). I am able to compile without any errors in x86_64.

alen12345 commented on 2015-08-26 21:25

@sachkris: I'm having the same problem, exactly the same compile error. @eleftg: I'm installing the package without modifications, therefore using the aur patch.

Can anyone please help me and @sachkris?

eleftg commented on 2015-04-12 23:46


you can have a log of the entire build process by typing:
makepkg 2>&1 | tee build.log

Can you please verify that during the configuration phase the following two lines show up?

-- Check for working CXX compiler: /usr/bin/mpicxx
-- Check for working CXX compiler: /usr/bin/mpicxx -- works

echo.motd commented on 2015-04-12 14:11

Scanning dependencies of target vtkIOInfovisPythonD
[ 78%] Building CXX object VTK/Wrapping/Python/CMakeFiles/vtkIOInfovisPythonD.dir/vtkISIReaderPython.cxx.o
/bin/sh: /usr/bin/mpicxx: No such file or directory
VTK/Wrapping/Python/CMakeFiles/vtkIOInfovisPythonD.dir/build.make:345: recipe for target 'VTK/Wrapping/Python/CMakeFiles/vtkIOInfovisPythonD.dir/vtkISIReaderPython.cxx.o' failed
make[2]: *** [VTK/Wrapping/Python/CMakeFiles/vtkIOInfovisPythonD.dir/vtkISIReaderPython.cxx.o] Error 127
CMakeFiles/Makefile2:20966: recipe for target 'VTK/Wrapping/Python/CMakeFiles/vtkIOInfovisPythonD.dir/all' failed
make[1]: *** [VTK/Wrapping/Python/CMakeFiles/vtkIOInfovisPythonD.dir/all] Error 2
Makefile:136: recipe for target 'all' failed
make: *** [all] Error 2
==> ERROR: A failure occurred in build().


[echo@hp ~]$ ls /usr/bin/mpicxx
[echo@hp ~]$


Please help with AUR paraview 4.3.1-1 installation.

eleftg commented on 2015-03-05 06:55


I'm afraid I can't help... Are you getting this error by applying this slightly different patch that you have mentioned below or by trying to build the package as provided here?

sachkris commented on 2015-03-05 06:19

I'm (still) getting this error, any idea how to fix it?

Building CXX object Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/paraDIS/paraDIS_lib/paradis.C.o
In file included from /usr/include/c++/4.9.2/vector:64:0,
from /usr/include/boost/format.hpp:17,
from /home/user/paraview/src/ParaView-v4.3.1-source/Utilities/VisItBridge/databases/paraDIS/paraDIS_lib/paradis.h:41,
from /home/user/paraview/src/ParaView-v4.3.1-source/Utilities/VisItBridge/databases/paraDIS/paraDIS_lib/paradis.C:6:
/usr/include/c++/4.9.2/bits/stl_vector.h: In instantiation of ‘void std::vector<_Tp, _Alloc>::_M_initialize_dispatch(_Integer, _Integer, std::__true_type) [with _Integer = int; _Tp = paraDIS::Arm*; _Alloc = std::allocator<paraDIS::Arm*>]’:
/usr/include/c++/4.9.2/bits/stl_vector.h:413:55: required from ‘std::vector<_Tp, _Alloc>::vector(_InputIterator, _InputIterator, const allocator_type&) [with _InputIterator = int; _Tp = paraDIS::Arm*; _Alloc = std::allocator<paraDIS::Arm*>; std::vector<_Tp, _Alloc>::allocator_type = std::allocator<paraDIS::Arm*>]’
/home/user/paraview/src/ParaView-v4.3.1-source/Utilities/VisItBridge/databases/paraDIS/paraDIS_lib/paradis.C:765:31: required from here
/usr/include/c++/4.9.2/bits/stl_vector.h:1252:59: error: invalid conversion from ‘int’ to ‘std::vector<paraDIS::Arm*>::value_type {aka paraDIS::Arm*}’ [-fpermissive]
_M_fill_initialize(static_cast<size_type>(__n), __value);
/usr/include/c++/4.9.2/bits/stl_vector.h:1298:7: note: initializing argument 2 of ‘void std::vector<_Tp, _Alloc>::_M_fill_initialize(std::vector<_Tp, _Alloc>::size_type, const value_type&) [with _Tp = paraDIS::Arm*; _Alloc = std::allocator<paraDIS::Arm*>; std::vector<_Tp, _Alloc>::size_type = unsigned int; std::vector<_Tp, _Alloc>::value_type = paraDIS::Arm*]’
_M_fill_initialize(size_type __n, const value_type& __value)
Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/build.make:5068: recipe for target 'Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/paraDIS/paraDIS_lib/paradis.C.o' failed
make[2]: *** [Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/paraDIS/paraDIS_lib/paradis.C.o] Error 1
CMakeFiles/Makefile2:12289: recipe for target 'Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/all' failed
make[1]: *** [Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/all] Error 2
Makefile:137: recipe for target 'all' failed

lahwaacz commented on 2015-03-03 07:51

@eleftg: Thanks, I missed the fact that ParaView does not use system libraries/packages but instead bundles the third-party software in the source tarball and the resulting package. In this case 'custom' is really appropriate...

eleftg commented on 2015-03-03 06:22

@lahwaacz: As indicated in the License_v1.2.txt file of the src directory, parts of ParaView are under the following licenses:

-- ParaView License v1.2 (Sandia Corporation / Kitware)
-- Kitware
-- Los Alamos National Laboratory
-- VTK license (BSD)
-- HDF5 license (BSD-like)
-- Xdmf License (BSD)
-- Protocol Buffer License (BSD)
-- NVD3 License (Apache)
-- D3 License (BSD)
-- jQuery License
-- jQuery UI License (MIT)

lahwaacz commented on 2015-03-02 18:57

Why is the license 'custom'? AFAIK it should be BSD:

eleftg commented on 2015-03-02 05:26

just adopted the package and uploaded PKGBUILD and patches for version 4.3.1

capitalaslash commented on 2015-01-29 16:15

i was able to compile 4.3.1 using patches from this PKGBUILD and the ones from the paraview-git aur package together.

sachkris commented on 2015-01-29 12:39

Im trying to compile version 4.3.1. Applied a patch for FreeTypeTools.cxx which is slightly different from the one from haraldkl (available here :

Now, Im not able to compile this file : Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/paraDIS/paraDIS_lib/paradis.C

error: invalid conversion from ‘int’ to ‘std::vector<paraDIS::Arm*>::value_type {aka paraDIS::Arm*}’

Anyone else having this problem?

danfunky commented on 2015-01-28 18:02

the new version 4.3.1 still requires the patch from haraldkl

--- a/VTK/Rendering/FreeType/vtkFreeTypeTools.cxx 2015-01-14 14:08:04.000000000 +0100
+++ b/VTK/Rendering/FreeType/vtkFreeTypeTools.cxx 2015-01-28 18:53:30.114241861 +0100
@@ -1198,7 +1198,7 @@
if (bitmap)
metaData.ascent = std::max(bitmapGlyph->top - 1, metaData.ascent);
- metaData.descent = std::min(-(bitmap->rows - (bitmapGlyph->top - 1)),
+ metaData.descent = std::min(-int(bitmap->rows - (bitmapGlyph->top - 1)),
@@ -2000,10 +2000,10 @@
fontSize, gindex, bitmapGlyph);
if (bitmap)
- bbox[0] = std::min(bbox[0], pen[0] + bitmapGlyph->left);
- bbox[1] = std::max(bbox[1], pen[0] + bitmapGlyph->left + bitmap->width);
- bbox[2] = std::min(bbox[2], pen[1] + bitmapGlyph->top - 1 - bitmap->rows);
- bbox[3] = std::max(bbox[3], pen[1] + bitmapGlyph->top - 1);
+ bbox[0] = std::min(bbox[0], int(pen[0] + bitmapGlyph->left));
+ bbox[1] = std::max(bbox[1], int(pen[0] + bitmapGlyph->left + bitmap->width));
+ bbox[2] = std::min(bbox[2], int(pen[1] + bitmapGlyph->top - 1 - bitmap->rows));
+ bbox[3] = std::max(bbox[3], int(pen[1] + bitmapGlyph->top - 1));

sachkris commented on 2015-01-26 04:02

Paraview v4.3.1 is out now

haraldkl commented on 2014-12-22 15:26

Sorry, the new package is actually

haraldkl commented on 2014-12-22 13:03

I never got an internal compiler error. In which file does this happen?

However, with gcc 4.9.2-1 I am now getting type mismatches and a failed build. I used the following patch to get rid of it:

--- a/VTK/Rendering/FreeType/vtkFreeTypeTools.cxx 2014-09-23 16:47:11.000000000 +0200
+++ b/VTK/Rendering/FreeType/vtkFreeTypeTools.cxx 2014-12-22 11:04:14.003492645 +0100
@@ -1183,7 +1183,7 @@
if (bitmap)
metaData.ascent = std::max(bitmapGlyph->top - 1, metaData.ascent);
- metaData.descent = std::min(-(bitmap->rows - (bitmapGlyph->top - 1)),
+ metaData.descent = std::min(-int((bitmap->rows - (bitmapGlyph->top - 1))),
@@ -1950,8 +1950,8 @@
if (bitmap)
bbox[0] = std::min(bbox[0], pen[0] + bitmapGlyph->left);
- bbox[1] = std::max(bbox[1], pen[0] + bitmapGlyph->left + bitmap->width);
- bbox[2] = std::min(bbox[2], pen[1] + bitmapGlyph->top - 1 - bitmap->rows);
+ bbox[1] = std::max(bbox[1], int(pen[0] + bitmapGlyph->left + bitmap->width));
+ bbox[2] = std::min(bbox[2], int(pen[1] + bitmapGlyph->top - 1 - bitmap->rows));
bbox[3] = std::max(bbox[3], pen[1] + bitmapGlyph->top - 1);

Most likely the simple casting here, is not the proper way to solve this, but I am not sure what the best solution would be.

I uploaded a package with this patch at

dcanelhas commented on 2014-11-06 17:03

I'm getting "internal compiler error: Segmentation fault"
(even when using haraldkl's patched version).

gcc 4.9.1-2
openmpi 1.8.3-1

haraldkl commented on 2014-10-16 07:45

The patch still seems to work, though its off by some lines here and there. My build succeeded now, and I put the source package with the patch from Debian for GCC 4.9 that is working for me at:

haraldkl commented on 2014-10-15 23:53

Yes, the error resulted in an aborted compilation, the complete makepkg output would be somewhat long. It still happens with 4.2, and the solution is patching the sources, such that GCC 4.9 is satisfied. I used the patch from the debian-science-commits list, posted before and got it working with that.

Kitpaddel commented on 2014-09-26 13:23

I get the same error as haraldkl. Was there a solution to this problem?

Full error:

Linking CXX shared library ../../../lib/
CMakeFiles/vtkFiltersParallelMPI.dir/vtkDistributedDataFilter.cxx.o: In function `vtkDistributedDataFilter::AddConstantUnsignedCharPointArray(vtkUnstructuredGrid*, char const*, unsigned char)':
vtkDistributedDataFilter.cxx:(.text+0x288b): warning: memset used with constant zero length parameter; this could be due to transposed parameters
collect2: Fehler: ld gab 1 als Ende-Status zurück
VTK/Filters/ParallelMPI/CMakeFiles/vtkFiltersParallelMPI.dir/build.make:232: recipe for target 'lib/' failed
make[2]: *** [lib/] Error 1
CMakeFiles/Makefile2:8675: recipe for target 'VTK/Filters/ParallelMPI/CMakeFiles/vtkFiltersParallelMPI.dir/all' failed
make[1]: *** [VTK/Filters/ParallelMPI/CMakeFiles/vtkFiltersParallelMPI.dir/all] Error 2
Makefile:137: recipe for target 'all' failed
make: *** [all] Error 2

matse commented on 2014-09-18 16:45

There is no owner and no status to this point on the table and I couldn't find a cmake option that enables me to compile it against system vtk libs - so no, at least for now :)

gdon commented on 2014-09-18 14:29

It seems it is now possible to use arch VTK precompiled libs with Paraview:

(last item in the table)

Would it be possible to update accordingly?

matse commented on 2014-09-16 17:55

Does the compilation process abort or are it just warnings like these? And if compilation fails, could you please provide the full output of the makepkg process?

haraldkl commented on 2014-09-16 14:45

I still get errors like this:
In function ‘void* memset(void*, int, size_t)’,
inlined from ‘static void vtkDistributedDataFilter::AddConstantUnsignedCharPointArray(vtkUnstructuredGrid*, const char*, unsigned char)’ at /tmp/yaourt-tmp-gk772/aur-paraview/src/ParaView-v4.2.0-RC1-source/VTK/Filters/ParallelMPI/vtkDistributedDataFilter.cxx:2810:29:
/usr/include/bits/string3.h:81:32: warning: call to ‘__warn_memset_zero_len’ declared with attribute warning: memset used with constant zero length parameter; this could be due to transposed parameters
__warn_memset_zero_len ();

matse commented on 2014-09-16 08:32

Just found the time to fix the PKGBUILD and uploaded paraview to the newest RC release available.

haraldkl commented on 2014-07-15 08:10

After updating to GCC 4.9, compilation of Paraview failed for me with a complaint about constant 0 argument in memset. I applied this patch: from the Debian package in the hope, that it fixes the problem.

GloW commented on 2014-05-23 09:05

Yes, you should compile paraview from source, it will take you 2/3 hours on standard computer.

The other way is to package your own version of openMPI, with a PKGBUILD and install it with pacman.

Another solution could be to remove openMPI dependency from aur paraview PKGBUILD, but i am not sure paraview will work with this solution.

I would go to first solution.

To continue futher discussion we should move to arch linux forum.

giacombum commented on 2014-05-23 08:44

What does it mean "You should go all the way and compile also paraview"? I've to compile paraview manually from paraview's sources?

Or there's a way to tell yourt that in the system there's already an openmpi version?

GloW commented on 2014-04-25 10:35

because it's the way packaging works ?...
Yaourt cannot know that you've installed your own version of openmpi.
You should go all the way and compile also paraview.

giacombum commented on 2014-04-25 10:24

I'm using modules and I've compiled from source openmpi with intel, gnu and open64 compilers: but after I load a module (for example openmpi-1.6.5-gnu) and I try to install paraview, yaourt says that I've to install openmpi. Any idea?

matse commented on 2014-04-11 09:05

Thanks for reporting the problem and already a fix here. I have changed the PKGBUILD as you proposed.

luismiguelgcg commented on 2014-04-11 07:19

I have a problem while compiling paraview:

MPI.c:(.text+0x6eb1): undefined reference to `dlopen'

It is solved with the following PKGBUILD and patch:


I've found the patch in the Paraview maillist:

luismiguelgcg commented on 2014-04-11 07:17

I have a problem while compiling paraview:

MPI.c:(.text+0x6eb1): undefined reference to `dlopen'

It is solved with the following PKGBUILD and patch:


TechLT commented on 2014-02-19 16:11

I got this error during upgrade:

Linking CXX shared library ../../../../lib/
[ 0%] Built target vtkmetaio
Makefile:136: recipe for target 'all' failed
make: *** [all] Error 2

I solved removing FREETYPE from line 44 of PKGBUILD

TechLT commented on 2014-02-18 16:12

I get this error during upgrade:

Linking CXX shared library ../../../../lib/
[ 0%] Built target vtksys
Linking CXX shared library ../../../../lib/
[ 0%] Built target vtkmetaio
Makefile:136: recipe for target 'all' failed
make: *** [all] Error 2

TechLT commented on 2014-02-18 16:12

Cannot updgrade:

Linking CXX shared library ../../../../lib/
[ 0%] Built target vtksys
Linking CXX shared library ../../../../lib/
[ 0%] Built target vtkmetaio
Makefile:136: recipe for target 'all' failed
make: *** [all] Error 2

GloW commented on 2014-01-14 10:48

v4.1 is out (for some time)

matse commented on 2013-12-05 08:42

Is your system fully up to date - which version of freetype2 are you using?
You are not supposed to have "/usr/include/ft2build.h" - which package does own this file?

backfist commented on 2013-12-03 14:46

/usr/include/ft2build.h:56:38: fatal error: freetype/config/ftheader.h: No such file or directory
#include <freetype/config/ftheader.h>
any ideas?

panda-z commented on 2013-10-22 11:09

since paraview was link agaist, and recently the hdf5 package upgrades from 1.8.10_patch1-1 to 1.8.11-1. (grep hdf5 /var/log/pacman.log)
So create a soft link /usr/lib/ points to /usr/lib/, and /usr/lib/ points to /usr/lib/ will do the trick.

Or you have to recompile paraview.

coincoin commented on 2013-10-22 10:48

Got this error on many computers since last pacman -Syu I "/usr/lib/paraview-4.0/paraview: error while loading shared libraries: cannot open shared object file: No such file or directory"

matse commented on 2013-10-18 10:09

This package should not need qt5-tools and qt5-xmlpatterns... I just compiled it twice without these packages (one time in a clean chroot and another time on my home computer). Cmake should link to the qt4 versions of all the needed tools (e.g. /usr/lib/qt4/bin/xmlpatterns).
Please pasetebin the complete output of "makepkg" and post it here, I will have a look at it - unfortunately I don't have much spare time at the moment and as you see my replies can take some time.

vitke commented on 2013-10-05 08:12

This package needs qt5-tools and qt5-xmlpatterns to compile. I found this out when the compilation failed after several hours at 98%. Please add these to the dependencies.

tapia commented on 2013-08-20 21:30

I thought that I had updated boost before I tried to compile (I tried to compile a couple of times with the same error). Well, now there is no problem any more. The problem must have been with boost.

[2013-08-18 15:13] [PACMAN] upgraded boost-libs (1.54.0-2 -> 1.54.0-3)
[2013-08-18 15:14] [PACMAN] upgraded boost (1.54.0-2 -> 1.54.0-3)

There are 4 hours of difference between my timezone and the one of this server, so I probably compiled and then updated the system and then reported the problem... strange that I did that, if I saw that boost was being updated.
Thanks anyway ;)

tapia commented on 2013-08-20 21:21

I thought that I had updated boost before I tried to compile (I tried to compile a couple of times with the same error). Well, now there is no problem any more. The problem must have been with boost.

[2013-08-18 15:13] [PACMAN] upgraded boost-libs (1.54.0-2 -> 1.54.0-3)
[2013-08-18 15:14] [PACMAN] upgraded boost (1.54.0-2 -> 1.54.0-3)

matse commented on 2013-08-20 17:56

No, I would say it's impossible that your error had anything to do with your language settings, are you sure that you didn't update your system afterwards (especially boost was just updated on the day when you posted your error here)?
Or just check compiling without the LC_all=... thing, I bet it will compile fine. You can also check your pacman log to see whether you updated your system after the compiling error - but nevertheless nice to hear that everything now works for you in the end!

tapia commented on 2013-08-20 00:07

Hi Matse,
sorry for the language problem. This is interesting, after I changed the language configuration with "LC_ALL=C", no problem occurred. Can it be that such a problem occurs because of the language configuration? I made no other change.

matse commented on 2013-08-19 15:20

Is your system fully up to date (especilly boost 1.54.0-3 and glibc 2.18-2)?
And please preppend "LC_ALL=C" before posting any output here, since probably more people speak English than Spanish (?)

matse commented on 2013-08-19 14:58

Is your system fully up to date (boost 1.54.0-3 and glibc 2.18-2)?
And please append "LC_ALL=C" before posting any output here, since probably more people speak English than Spanish (?)

tapia commented on 2013-08-18 19:21

while trying to compile this, i got this error:

[ 13%] Building CXX object Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/vtkVisItSASReader.cxx.o
In file included from /home/tapia/Descargas/AUR/paraview/src/build/Utilities/VisItBridge/databases/vtkVisItSASReader.cxx:38:0:
/home/tapia/Descargas/AUR/paraview/src/ParaView-v4.0.1-source/Utilities/VisItBridge/databases/SAS/avtSASFileFormat.h:59:14: error: ‘int64_t’ ya se declaró en este ámbito
using boost::int64_t;
make[2]: *** [Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/vtkVisItSASReader.cxx.o] Error 1
make[1]: *** [Utilities/VisItBridge/databases/CMakeFiles/vtkIOVisItBridge.dir/all] Error 2
make: *** [all] Error 2
==> ERROR: Se produjo un error en build().

Any idea what could this be?

matse commented on 2013-08-17 09:17

hamid: Thanks for submitting this bug report! And yes, my link pointed to the homepage of the bugtracker because I reported the issue on the paraview mailing list and was asked to open a bug report.

Anonymous comment on 2013-08-11 08:43

@matse: I have reported the issue in their bug tracker (#0014215) that day, but it has not been assigned yet and no one have replied to me. May be I have not done it the right way, or may be supporting ffmpeg 2.0 is not urgent for them. Since the bug is so local (only one place in the code that needs to be corrected), I suppose applying it in the code would be quite easy for them.
PS. Your link points to the homepage of bugtracker and not the issue you opened
PPS. As I was finishing this, I had a second look and it appears as resolved.

matse commented on 2013-08-09 11:54

Hey hamid,
thanks for finding the issue and providing a fix.
Did you already report this to upstream, since I reported the issue some days ago and a dev asked me to open a bugreport on

Since I don't have regularly internet access at the moment it would be nice if you could do this. I will stick to ffmpeg-compat here until the problem is fixed upstream, greetings

Anonymous comment on 2013-08-07 06:37

The latest ffmpeg has changed CodecID to AVCodecID. You can make paraview compile against ffmpeg 2.0 by adding:
sed -i 's/CodecID/AVCodecID/g' ./VTK/IO/FFMPEG/vtkFFMPEGWriter.cxx
to the PKGBUILD, as I did in my copy. I will report this to the upstream but until it gets fixed, this can do the trick.

Raptorista commented on 2013-08-02 09:24

Yess, it works! And you rock ;)

matse commented on 2013-08-02 01:33

Raptorista: You are right, paraview doesn't compile against ffmpeg 2.0 yet, probably an API breakage. I changed the PKGBUILD so that paraview now is linked against ffmpeg-compat, which works fine here. Happy recompiling!

Raptorista commented on 2013-08-01 17:46

Today's upgrade of ffmpeg broke paraview. Launching from terminal complains "paraview: error while loading shared libraries: cannot open shared object file: No such file or directory".
Downgrading ffmpeg and ffmpeg-compat makes the missing library change to libx264, confirming paraview is looking for files that have been moved.

matse commented on 2013-06-25 04:28

Hell-G: Hi, hmm these packages should not be neccessary, since paravie uses qt4 anyway (and I was able built this version in a clean chroot without these packages). Do you use my plain "vanilla" PKGBUILD and can you pastebin the "configure" part of the output when you run makepkgg (so just the first few hundred lines...)?

Hell-G commented on 2013-06-24 19:14


Thank you for the package! I had to manually install the packages qt5-xmlpatterns and qt5-tools (for qhelpgenerator) in order to successfully build the package with makepkg (makepkg -s didn't solve the problem of the dependencies). I think these were necessary to build the documentation.
Are these missing dependencies or is my system just weird?


matse commented on 2013-06-23 15:17

Raptorista: Nice that paraview works for you now, too... and no, there is no need to pastebin the output of the working makepkg command, maybe we will never get to know what went wrong last time - but who cares ;)

Raptorista commented on 2013-06-20 11:12

Mmm, this is really weird!
I downloaded your PKGBUILD and ran makepkg and it just compiled fine...
makepkg output looks pretty normal, I haven't "pastebined" it because it exceeds size limit, but can find a way if you still want it.

matse commented on 2013-06-19 20:24

Would you please also post the full output of the makepkg command with my PKGBUILD? And does removing line 50 and 51 in the PKGBUILD solve the problem for you?

Raptorista commented on 2013-06-19 19:46

I download the sources from the website, untar, (mkdir && cd) build, cmake .. and the result is

matse commented on 2013-06-19 17:38

Raptorista: can you also pastebin the output of cmake? Best post it somewhere else like since it's probably a bunch of lines...

Raptorista commented on 2013-06-19 16:25

@matse: Thanks for supporting!
My system is up-to-date, here is

$ pacman -Qm
atlas-lapack 3.10.1-1
dropbox 2.2.3-1
firefox-oxygen-kde 4.0_b2-2
firefox-raismth 3.0.4-3
gcc-gcj 4.8.0-2
gcc-gcj-ecj 4.5-2
gnuplot-py 1.8-3
knights 2.5.0-2
lib32-dbus-core 1.6.10-1
libcgns-paraview 3.1.4-1
metis 5.0.2p3-1
package-query 1.2-2
parmetis 4.0.2-2
pdftk 1.45-2
psmouse-alps-driver 2:1.3-4
ptscotch-openmpi 6.0.0-2
pyqt 4.10.1-1
pyqt-common 4.10.1-1
pyqwt 5.2.0-8
python2-h5py 2.1.2-1
python2-pyqt 4.10.1-1
python3-h5py 2.1.2-1
qwt5 5.2.3-1
tor-browser-en 2.3.25_8-1
yaourt 1.3-1

Besides, you might wanna see
$ gcc -v
Using built-in specs.
Target: x86_64-unknown-linux-gnu
[... Stuff...]
gcc version 4.8.1 (GCC)

Hope it helps!

matse commented on 2013-06-19 15:31

karlmarx80: Yes, upstream just uploaded a new source tar.gz today that has a different checksum as the one from the 06/14/13. I updated the PKGBUILD, and this should work now.

Anonymous comment on 2013-06-19 15:16


I encoutered a problem while trying to install paraview 4.0.1-1.

It seems that ParaView-v4.0.1-source.tgz can not be validated.... Do you have any idea on how to correct the problem?

Here is the error I get:

> makepkg
==> Making package: paraview 4.0.1-1 (mer jun 19 17:03:33 CEST 2013)
==> Checking runtime dependencies...
==> Checking buildtime dependencies...
==> Retrieving sources...
-> Downloading ParaView-v4.0.1-source.tgz...
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 45.6M 100 45.6M 0 0 632k 0 0:01:13 0:01:13 --:--:-- 560k
-> Found paraview.png
-> Found paraview.desktop
==> Validating source files with sha1sums...
ParaView-v4.0.1-source.tgz ... FAILED
paraview.png ... Passed
paraview.desktop ... Passed
==> ERROR: One or more files did not pass the validity check!

matse commented on 2013-06-19 14:57

Raptorista: Is your system completely up to date? This message indicates that you use a pretty old version of gcc, you should have 4.8.1. If your system is up to date pastebin the output of "pacman -Qm".

Raptorista commented on 2013-06-18 10:14

When I try to compile, my Arch complains
-- Detecting Fortran/C Interface
-- Detecting Fortran/C Interface - Found GLOBAL and MODULE mangling
CMake Error at VTK/CMake/GenerateExportHeader.cmake:177 (if):
if given arguments:


Unknown arguments specified
Call Stack (most recent call first):
VTK/CMake/GenerateExportHeader.cmake:352 (_test_compiler_hidden_visibility)
VTK/CMake/vtkModuleMacros.cmake:521 (generate_export_header)
Utilities/PythonInitializer/CMakeLists.txt:27 (vtk_module_library)

-- Configuring incomplete, errors occurred!

What should I do? :S

matse commented on 2013-04-27 12:14

the python libraries in /usr/lib/paraview... are the place where the devs want to put them. If you want to use the libraries outside paraview's python shell the devs want you to edit yout PYTONPATH variable, see

So I personally don't see the need to manually move the libs to /usr/lib/python... whatever or create symlinks there (I guess as the debian people do?).

Further upstream doesn't want you to link paraview against you system mpi4py library, it seems that there was once a cmake option "PARAVIEW_USE_SYSTEM_MPI4PY" but it's not available in the latest paraview releases anymore and the devs ship their own mpi4py lib with it. You might ask upstream for the cause.

mauro2 commented on 2013-04-22 21:05

The python2 libraries end up in the wrong place with this build, namely in
/usr/lib/paraview-*/site-packages with some scattered *.so files in /usr/lib/paraview-*. n Whereas they should be, I think, in /usr/lib/python2.7/dist-packages/. But just moving them might impact the paraview installation...

Also, the mpi4py probably should be a dependency rather than being installed with paraview.

Other than that, all sweetly working.

matse commented on 2013-02-22 10:32

Yes you have to install python2-numpy in order to use the python-calculator.
Just install the package and it should work, I have uploaded a new release with python2-numpy as optional dependency.

Anonymous comment on 2013-02-22 09:42

Does anyone also have a problem using the Python Calculator? I get the following error when I try to use it:

Traceback (most recent call last):
File "<string>", line 4, in <module>
File "/usr/lib/paraview-3.98/site-packages/paraview/vtk/", line 5, in <module>
sure that it is installed properly.")
RuntimeError: This module depends on the numpy module. Please makesure that it is installed properly.

Moreover, when I try to import numpy from the python shell I get:

>>> import numpy
Traceback (most recent call last):
File "<console>", line 1, in <module>
ImportError: No module named numpy

Does someone knows how to solve this problem?

Thanks very much.

mfs commented on 2013-02-04 08:10

matse: Thanks for including the lines. I'm using openmpi 1.6.3-1 from [extra], so that should not be the reason for the problem. But I don't care finding the root cause too much, as it is working now ;-)

matse commented on 2013-02-03 15:36

Trigger: You are right, boost was missing, added it as dependency and increased release number. Thanks for reporting and I should not be too lazy to build in a chroot environment in the future...

matse commented on 2013-02-03 14:24

mfs: Thanks for reporting your issue, I have changed the PKGBUILD without changing version number - even if I couldn't reproduce your error.
Do you by any chance use another implementation of MPI then OpenMPI (from the official repos)?

mfs commented on 2013-02-01 19:44

I've solved the linking issue by adding


to the cmake flags in PKGBUILD. This sets the compiler to wrappers for gcc and g++ provided by openmpi, which is the recommended way of compiling MPI applications according to So I suggest to add these two lines to your PKGBUILD.

mfs commented on 2013-02-01 18:06

I'd love to use this program, but compiling always fails at 5%:

[ 5%] Building CXX object VTK/ThirdParty/xdmf2/vtkxdmf2/libsrc/CMakeFiles/vtkxdmf2.dir/XdmfDsmMsg.cxx.o
Linking CXX shared library ../../../../../lib/
CMakeFiles/vtkxdmf2.dir/XdmfArray.cxx.o: In function `MPI::Win::Set_name(char const*)':
XdmfArray.cxx:(.text._ZN3MPI3Win8Set_nameEPKc[_ZN3MPI3Win8Set_nameEPKc]+0x5): undefined reference to `MPI_Win_set_name'
CMakeFiles/vtkxdmf2.dir/XdmfArray.cxx.o: In function `MPI::Win::Set_attr(int, void const*)':
XdmfArray.cxx:(.text._ZN3MPI3Win8Set_attrEiPKv[_ZN3MPI3Win8Set_attrEiPKv]+0x5): undefined reference to `MPI_Win_set_attr'

Lots of similar messages follow and compiling eventually fails. The full output of makepkg is here:

I've already tried to toggle PARAVIEW_USE_MPI from ON to OFF, but to no avail.

Do you have an idea what could be causing the issue?

Anonymous comment on 2013-01-30 10:59

Thank you for providing this package here. Compilation went fine for me and did not take too long on an i7. I had to additionally install boost.