Unable to load the shared library about datoviz HOT 7 CLOSED

I think the key bit is:

ImportError: /lib/x86_64-linux-gnu/librsvg-2.so.2: undefined symbol: cairo_tag_end

First, what is your exact operating system?

from datoviz.

Also, what is your Python environment?

from datoviz.

I'm on Ubuntu 20.04.2 LTS and use Anaconda 4.9.2 with python 3.7.9

from datoviz.

Okay, I think this is the issue: https://stackoverflow.com/a/63305833/1595060

If I understand correctly, Anaconda overrides a system lib (librsvg) with its own, and the librsvg provided in Ubuntu 20.04 (used when compiling Datoviz) doesn't match the one provided by Anaconda (used when running Python). So you have to tell Python to use the system librsvg instead of the Anaconda one. The link above should provide some guidance, I'll try to find a more automated way to fix this.

from datoviz.

For the record, it seems this library is one of the many dependencies brought by ffmpeg when Datoviz is compiled with ffmpeg support (ffmpeg is only used by the integrated live screencast video feature, but it may also be used in the future if we ever implement a Video visual).

So, you could also go in the CMakeLists.txt and disable compilation with ffmpeg by replacing option(DATOVIZ_WITH_FFMPEG "Build Datoviz with FFMPEG support" ON) by option(DATOVIZ_WITH_FFMPEG "Build Datoviz with FFMPEG support" OFF). This will disable the screencast feature, but will also remove the broken librsvg dependency.

For future reference, here are the dependencies of Datoviz when ffmpeg support is OFF (Ubuntu 20.04):

$ ldd build/libdatoviz.so
	linux-vdso.so.1 (0x00007ffdba957000)
	libglfw.so.3 => /lib/x86_64-linux-gnu/libglfw.so.3 (0x00007efc65bc4000)
	libvulkan.so.1 => /home/cyrille/vulkan/x86_64/lib/libvulkan.so.1 (0x00007efc6596f000)
	libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007efc6594c000)
	libpng16.so.16 => /lib/x86_64-linux-gnu/libpng16.so.16 (0x00007efc65914000)
	libstdc++.so.6 => /lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007efc65733000)
	libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007efc655e2000)
	libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007efc655c7000)
	libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007efc653d5000)
	/lib64/ld-linux-x86-64.so.2 (0x00007efc66024000)
	libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007efc653cf000)
	libX11.so.6 => /lib/x86_64-linux-gnu/libX11.so.6 (0x00007efc65292000)
	libz.so.1 => /lib/x86_64-linux-gnu/libz.so.1 (0x00007efc65276000)
	libxcb.so.1 => /lib/x86_64-linux-gnu/libxcb.so.1 (0x00007efc6524a000)
	libXau.so.6 => /lib/x86_64-linux-gnu/libXau.so.6 (0x00007efc65244000)
	libXdmcp.so.6 => /lib/x86_64-linux-gnu/libXdmcp.so.6 (0x00007efc6523c000)
	libbsd.so.0 => /lib/x86_64-linux-gnu/libbsd.so.0 (0x00007efc65222000)

Note for future package managers: if creating a conda package, I guess you should make sure that you use the Anaconda ffmpeg and other libraries instead of the system libraries.

from datoviz.

Thanks! Running the python script as follows works for me:
LD_PRELOAD=/lib/x86_64-linux-gnu/libcairo.so.2 python bindings/cython/examples/quickstart.py

from datoviz.

Great, thanks! Will update the install instructions for Anaconda users on Ubuntu.

from datoviz.