This document describes how to install SUMO on Linux from sources. If you don't want to extend SUMO, but merely use it, you might want to download one of our pre-built binary packages instead.
To be able to run SUMO on Linux, just follow these steps:
- Install all of the required tools and libraries
- Get the source code
- Build the SUMO binaries
For ubuntu this boils down to
sudo apt-get install git cmake python3 g++ libxerces-c-dev libfox-1.6-dev libgdal-dev libproj-dev libgl2ps-dev python3-dev swig default-jdk maven libeigen3-dev
git clone --recursive https://github.com/eclipse-sumo/sumo
cd sumo
export SUMO_HOME="$PWD"
cmake -B build .
cmake --build build -j$(nproc)
Note
Do not build in an active Anaconda environment! The libraries have sometimes different versions than Anaconda which will break the build or the executables.
Each of these steps is described in more detail and with possible alternatives below.
Installing required tools and libraries#
- For the build infrastructure you will need cmake together with a moderately recent g++ (4.8 will do) or clang++ (or any other C++11 enabled compiler).
- The library Xerces-C is always needed. To use sumo-gui you also need Fox Toolkit in version 1.6.x. It is highly recommended to also install Proj to have support for geo-conversion and referencing. Another common requirement is network import from shapefile (arcgis). This requires the GDAL library. To compile you will need the devel versions of all packages. For openSUSE this means installing libxerces-c-devel, libproj-devel, libgdal-devel, and fox16-devel. For ubuntu the call is above. The installation of swig, python3-dev and the jdk enables also the build of libsumo while eigen3 is necessary for the overheadwire model. There are some outdated platform specific and manual build instructions for the libraries
- Optionally you may want to add
- ccache (to speed up builds)
- ffmpeg-devel (for video output),
- libOpenSceneGraph-devel (for the experimental 3D GUI),
- gtest (for unit testing, do not use 1.13 or later)
- gettext (for internationalization)
- texttest, xvfb and tkdiff (for the acceptance tests, include python3-gobject for the texttest GUI)
- flake, astyle and autopep for style checking
- see also further dependencies for GUI testing
The package names above are for openSUSE, for ubuntu the call to get the most important optional libraries and tools is:
sudo apt-get install ccache libavformat-dev libswscale-dev libopenscenegraph-dev python3-pip python3-build
sudo apt-get install libgtest-dev gettext tkdiff xvfb flake8 astyle python3-autopep8 python3-gi-cairo gir1.2-gtk-3.0
sudo apt-get install python3-pyproj python3-rtree python3-pandas python3-pulp python3-ezdxf
python3 -m pip install texttest
For the Python tools there are some more requirements depending on which tools you want to use. If you want to install
everything using pip do python3 -m pip install -r tools/requirements.txt -r tools/req_dev.txt
.
Getting the source code#
For the correct setting of SUMO_HOME you have to remember the correct path, where you build your SUMO, the SUMO build path. This path is shown with pwd at the end of getting the source code. If you want to develop actively on sumo we strongly recommend to use the git repository. Please build the tarball version if you need for some reason a specific version of sumo.
repository checkout (recommended)#
The following commands should be issued:
git clone --recursive https://github.com/eclipse-sumo/sumo
cd sumo
git fetch origin refs/replace/*:refs/replace/*
pwd
The additional fetch of the replacements is necessary to get a full local project history.
release version or nightly tarball#
Download sumo-src-1.21.0.tar.gz or https://sumo.dlr.de/daily/sumo-src-git.tar.gz
tar xzf sumo-src-<version>.tar.gz
cd sumo-<version>/
pwd
Definition of SUMO_HOME#
Before compiling is advisable to define the environment variable SUMO_HOME. SUMO_HOME must be set to the SUMO build path from the previous step. Assuming that you placed SUMO in the folder "/home/<user>/sumo-<version>", if you want to define only for the current session, type in the console
export SUMO_HOME="/home/<user>/sumo-<version>"
If you want to define for all sessions (i.e. for every time that you run your Linux distribution), go to your HOME folder, and find one of the next three files (depending of your Linux distribution): .bash_profile, .bash_login or .profile (Note that these files can be hidden). Then edit the file, add the line from above at the end and restart your session.
You can check that SUMO_HOME was successfully set if you type
echo $SUMO_HOME
and console shows "/home/<user>/sumo-<version>"
Installing Python packages for the tools#
Calling the tools from netedit requires a list of Python packages to generate templates during compilation. Many of them might be available with the package manager of your distribution and most of the time we prefer to use those. For ubuntu this currently means, you should first do
sudo apt-get install python3-pyproj python3-rtree python3-pandas flake8 python3-autopep8 python3-pulp python3-ezdxf
and then install the remaining parts using pip:
python3 -m pip install -r tools/requirements.txt -r tools/req_dev.txt
The pip installation will ensure that all libraries are there, so it is safe to skip the first apt-get
step.
If you need information about the minimum required versions of the packages read them directly
from the requirements.txt. Be aware that
the minimum versions in the requirements file just reflect our current test server setup, so you might also get away with earlier versions.
If you want to reproduce our test server setup exactly, then use the versions in
req_test_server.txt
Unfortunately some old pip packages are incomplete, so this requires sudo apt install libspatialindex-dev
.
Building the SUMO binaries with cmake#
To build with cmake version 3.5 or higher is required.
Create a build folder for cmake (in the SUMO root folder) and configure SUMO with the full set of available options like GDAL and OpenSceneGraph support (if the libraries are installed):
cmake -B build .
to build the debug version just use
cmake -D CMAKE_BUILD_TYPE=Debug -B build .
Note
On some platforms the required cmake executable is called cmake3.
Other useful cmake configuration options:
-D PROFILING=ON
enable profiling instrumentation for gprof (gcc build only)-D COVERAGE=ON
enable coverage instrumentation for lcov (gcc build only)-D CHECK_OPTIONAL_LIBS=OFF
disable all optional libraries (only include EPL compatible licensed code)-D CMAKE_BUILD_TYPE=RelWithDebInfo
enable debug symbols for debugging the release build or using a different profiler-D PROJ_LIBRARY=
disable PROJ-D FOX_CONFIG=
disable FOX toolkit (GUI and multithreading)-D PYTHON_EXECUTABLE=/usr/bin/python3
select a different python version (also for libsumo / libtraci)-D MVN_EXECUTABLE=
disable maven packaging (especially useful if you have no network connection)-D ENABLE_CS_BINDINGS=ON
enable C# bindings when compiling libsumo / libtraci
After this is finished, run
cmake --build build -j $(nproc)
The nproc
command gives you the number of logical cores on your
computer, so that make will start parallel build jobs which makes the
build a lot faster. If nproc
is not available on your system, insert a
fixed number here or leave the option out. You may also try
cmake --build build -j $(grep -c ^processor /proc/cpuinfo)
Building with clang#
If you want to use a different compiler (just for the fun of it or because it has additional features) you can enable it at configure time. Our current clang configuration for additional static code checking enables the following CXXFLAGS:
-stdlib=libstdc++ -fsanitize=undefined,address,integer -fno-omit-frame-pointer -fsanitize-blacklist=$SUMO_HOME/build_config/clang_sanitize_blacklist.txt
You may of course leave out all the sanitizer-checks you don't want but the stdlib option has to be set. The blacklist filters out a known bug in the cstdlib. For details see the clang documentation.
The current CMake configuration already contains this for the debug build, so for building with CMake and clang just change to your build dir and use
CXX=clang++ cmake -DCMAKE_BUILD_TYPE=Debug --build build -j $(nproc)
The clang-debug-build will detect memory leaks (among other things) If the errors are reported with cryptic hexadecimal numbers as
Indirect leak of 72 byte(s) in 1 object(s) allocated from:
#0 0xa4ee2d (.../sumo/bin/netconvertD+0xa4ee2d)
set the following environment variable to point to the llvm-symbolizer executable:
export ASAN_SYMBOLIZER_PATH=/usr/bin/llvm-symbolizer
before running the executable.
Please note that the undefined behavior checker is very sensitive and will report
some false negatives also in system libraries. It is recommended to use
export UBSAN_OPTIONS=suppressions=$SUMO_HOME/build_config/clang_ubsan_suppressions.txt
before calling the executable.
Installing the SUMO binaries#
This (optional) step will copy the SUMO binaries to another path, so that you can delete all source and intermediate files afterwards. If you do not want (or need) to do that, you can simply skip this step and run SUMO from the bin subfolder (bin/sumo-gui and bin/sumo).
If you want to install the SUMO binaries into your system, run
sudo cmake --install build
You have to adjust your SUMO_HOME variable to the install dir (usually /usr/local/share/sumo)
export SUMO_HOME=/usr/local/share/sumo
Uninstalling#
CMake provides no make uninstall
so if you ever want to uninstall, run
sudo xargs rm < install_manifest.txt
from the same folder you ran make install
. This will leave some empty
directories, so if you want to remove them as well, double check that
$SUMO_HOME points to the right directory (see above) and run
sudo xargs rm -r $SUMO_HOME
Building Python wheels for sumolib, traci and libsumo#
If you want to distribute sumolib, traci and/or libsumo as wheels you can build those wheels directly from the tools tree. Please be aware that nightly builds of those packages are also available on https://test.pypi.org
pip install wheel build
cd tools
python build_config/version.py build_config/setup-sumolib.py ./setup.py
python -m build --wheel
python build_config/version.py build_config/setup-traci.py ./setup.py
python -m build --wheel
python build_config/version.py build_config/setup-libsumo.py ./setup.py
python -m build --wheel
You will need a recent version of pip (>=22) for this to work. If for some reason you cannot update your pip you can also use the (discouraged!) method of calling setup.py directly.
cd tools
python build_config/setup-sumolib.py bdist_wheel
Please note that you always need to be in the tools directory for this to work and your wheels will be placed in tools/dist. Furthermore the traci and the sumolib wheel are platform and Python version independent while libsumo depends on the exact platform and Python you built it with.
(Frequent) Rebuilds#
If you did a repository clone you can simply update it by doing git pull
from inside the SUMO_HOME folder. Then change to the build directory and run
make -j $(nproc)
again.
If your underlying system changed (updated libraries) or you experience other build problems please try a clean build first by removing the build directory (or at least the CMakeCache.txt) and running cmake and make again before reporting a bug.
If you find yourself building very often after minor changes, consider installing ccache and run cmake again. It will be picked up automatically and can dramatically improve build speed.
How to build JuPedSim and then build SUMO with JuPedSim#
In this section, you will learn how to build the latest version of the pedestrian simulator JuPedSim and how to compile SUMO with this latest version of JuPedSim. First of all, clone the JuPedSim repository:
git clone https://github.com/PedestrianDynamics/jupedsim
Note that this will clone the full repository, including the latest version of JuPedSim. We strongly recommend to build the latest release of JuPedSim (not the master branch), which is officially supported by SUMO. You can consult the JuPedSim build procedure; hereafter we propose a similar procedure. First check which is the latest release then in the cloned directory checkout to the latest release and do a regular cmake build. For example, for JuPedSim release v1.2.1, you would need to type:
cd jupedsim
git checkout v1.2.1
cmake -B build .
cmake --build build
sudo cmake --install build
Now you should make sure GEOS is installed (sudo apt-get install libgeos-dev
) and
continue with the standard build procedure above.
Tweaking the JuPedSim build#
If you do not want to install jupedsim into your system, you can specify an alternative install directory like this:
cmake -B build -DCMAKE_INSTALL_PREFIX=$PWD/../jupedsim-install .
cmake --build build
cmake --install build
This will install jupedsim in the directory jupedsim-install
right beside the checkout. If you installed jupedsim
into the system or in a jupedsim-install
directory beside sumo, the standard cmake call of SUMO will find it
automatically.
Please be aware that if you want to install sumo into the system, you also need to install jupedsim into the system.
To tweak or debug the jupedsim build you can also change the configuration to Debug (with -DCMAKE_BUILD_TYPE=Debug
)
and also enable multithreading (with -j4
) as usual with CMake. If you have different jupedsim versions or choose a
different install path, you can notify CMake where JuPedSim is installed by setting JUPEDSIM_CUSTOMDIR
when calling CMake.
For further remarks on the use of JuPedSim inside SUMO, please consult the documentation on the model.
Troubleshooting#
Problems with the socket subsystem#
Problem:
recv ./foreign/tcpip/libtcpip.a(socket.o) (symbol belongs to implicit dependency /usr/lib/libsocket.so.1)
Solution: https://lists.danga.com/pipermail/memcached/2005-September/001611.html
ld cannot find an existing library (Fedora-23)#
Problem:
/usr/bin/ld: cannot find -lfreetype
ls -lah /usr/lib64/libfreetype*
lrwxrwxrwx. 1 root root 21 Jul 28 15:54 /usr/lib64/libfreetype.so.6 -> libfreetype.so.6.12.0
lrwxr-xr-x. 1 root root 689K Jul 28 15:54 /usr/lib64/libfreetype.so.6.12.0
Solution: Install the dev package; for fedora:
sudo yum install freetype-devel
For details see stackoverflow discussion.
ld cannot find certain functions in an existing external library#
Make sure you don't have an environment like anaconda installed which modifies your library search path.
Additional notes for Cygwin users#
At the moment GUI building is still troublesome. It depends whether you want to use the X-Server or native Windows GUI. We tried native Windows GUI only and had to change the installed libFOX-1.4.la such that it contains
dependency_libs=' -lgdi32 -lglaux -ldl -lcomctl32 -lwsock32 -lwinspool -lmpr
-lpthread -lpng /usr/lib/libtiff.la /usr/lib/libjpeg.la -lz -lbz2 -lopengl32 -lglu32'
Your mileage may vary.