Cannot compile Paraview 5.4.1

ParaView Support
1

Installing Paraview unsing paraview-superbuild, I’m facing with:

[ 78%] Completed 'osmesa'
[ 78%] Built target osmesa
Makefile:129: recipe for target 'all' failed

My ccmake config is:

CMAKE_INSTALL_PREFIX=/opt/paraview-5.4.1
ENABLE_ffmpeg=ON
ENABLE_fontconfig=ON
ENABLE_freetype=ON
ENABLE_matplotlib=ON
ENABLE_mpi=ON
ENABLE_numpy=ON
ENABLE_osmesa=ON
ENABLE_ospray=ON
ENABLE_png=ON
ENABLE_python=ON
ENABLE_pythonsetuptools=ON
ENABLE_scipy=ON
PYTHON_EXECUTABLE:FILEPATH=/usr/bin/python2
PYTHON_INCLUDE_DIR:PATH=/usr/include/python2.7
PYTHON_LIBRARY:FILEPATH=/usr/lib/x86_64-linux-gnu/libpython2.7.so

, Since I’m following this.

sys.maxunicode>65535 is set to TRUE.

Sometimes the same error happens with vtkm:

[ 63%] Completed 'vtkm'
[ 63%] Built target vtkm
Makefile:129: recipe for target 'all' failed
make: *** [all] Error 2

I’m bad at CMake, please, give me some clue what this error could be about?
I want to use Paraview for visualize a result from libpointmatcher (ICP library for 2-D/3-D mapping in Robotics)
Thank you!

2

Are you sure you are building from the v5.4.1 tag in the superbuild? Actually, that may not be helpful for figuring out what the errors you are seeing mean, but it is a good thing to check nonetheless.

Does the build complete if you run make a second time?

3

HI, thanks for your response
Yes, I’m building from the v5.4.1 tag.
No, I’ve tried multiple times, that’s basically the error, while running make second time.

[ 63%] Completed 'vtkm'
[ 63%] Built target vtkm
Makefile:129: recipe for target 'all' failed
make: *** [all] Error 2
4

Could it be some python conflict error?
I did comment ROS source lines in ~/.bashrc , if related.

5

I don’t suspect a Python error. @ben.boeckel have we ever seen this?

What command are you running to execute the build? make should be sufficient without the -j option - parallel building happens at the project level. Maybe if you are including -j that is leading to a problem (wild guess, haven’t tried it).

6

Strangely enough, make without the -j option was the way to go.

I’d also like to point out, due to slow internet connection, I’ve been pre-downloading required dependencies into the CMAKE_INSTALL_PREFIX/downloads/ dir., where it is expected to be found by superbuild.

So, reason could be wrongly set permission access , although it was set to 777 or broken files.
Is there any way to check whether compressed files are corrupt or not?

Anyways, installing without -j flag and without pre-downloading archives solved the problem.

Thank you, @cory.quammen, for your help!

7

@cory.quammen @artemiialessandrini Surprisingly, I have seen issues with a -j before, but not for a long time. When you do get a good build, you may want to also try building with Paraview 5.7. 5.4.1 was a really good, stable release, but it is a few years old at this point.

8

Thanks for your comment, @wascott

Currently I’m using it only as a visualization tool for 2D PointClouds, which seems to be quite an easy task.

When digging deeper, I’ll definitely give it a try!

9

Don’t forget that there is a really easy solution - Kitware has binaries on the paraview.org website. If you could use these…

10

I imagine that there could certainly be parallel build issues in 5.4.1 with Makefiles. I believe Ninja was the vastly preferred method at the time, but the superbuild may have not supported Ninja then either. It’s hard to tell without the output of the actual failure (which in a parallel build may be far away from the make: *** [all] Error 2 output you have above).