This is a meta repository (so-called "superbuild") that uses CMake and YCM to automatically download and compile software developed in the robotology GitHub organization, such as the YARP middleware or software used to run the iCub humanoid robot.
CMake is an open-source, cross-platform family of tools designed to build, test and package software.
A YCM Superbuild is a CMake project whose only goal is to download and build several other projects.
If you are familiar with ROS, it is something similar to catkin or colcon workspace, but using pure CMake for portability reasons and for customizing the build via CMake options. Furthermore, the robotology-superbuild also contains some infrastructure to build binaries of the contained projects for some platforms.
You can read more about the superbuild concept in YCM documentation or in the related IRC paper.
The robotology-superbuild is an infrastructure to simplify development and use of open source research software developed at the Italian Institute of Technology, in particular as part of the iCub project.
As a huge number of software projects are contained in the robotology-superbuild, and a tipical user is only interested in some of them, there are several options to instruct the superbuild on which packages should be built and which one should not be built. In particular, the robotology-superbuild is divided in different profiles, that specify the specific subset of robotology packages to build. You can read more on the available profiles and how to enable them in the doc/cmake-options.md#profile-specific-documentation.
Furthermore, some dependencies of software contained in the robotology-superbuild are either tricky to install or proprietary, and for this reason software that depends on those optional dependencies can be enabled or disabled with specific options,as documented in doc/cmake-options.md#dependencies-specific-documentation.
For what regards versioning, software in the robotology-superbuild can be consumed in two forms:
In this form, the superbuild will get the latest changes for a branch of each subproject, and will build it. This has the advantage that you get all the latest changes from the software contained in the robotology-superbuild, while the downside that the specific software that you use may change at each update. The rolling update can be used only when building robotology-superbuild software from source. By default, the robotology-superbuild uses the latest "stable" branches of the robotology repositories, but in some cases it may be necessary to use the "unstable" active development branches. For this advanced functionalities, please refer to the documentation on changing the default project tags, available at doc/change-project-tags.md.
Once every three months, a set of releases of the software in the robotology-superbuild is freezed and used as a "Distro Release", following the policies of iCub software described in https://icub-tech-iit.github.io/documentation/sw_versioning_table/ . Releases can be used both when building the software from source, and when obtaining it from binaries.
The available releases can be seen on GitHub's release page.
We provide binary packages for Linux, macOS and Windows of the software contained in the robotology-superbuild via the conda package manager, relying on the community-mantained conda-forge channel and for some packages on our own robotology conda channel.
Please refer to doc/conda-forge.md document for instructions on how to install the conda binary packages, in particular the Binary Installation section.
Note that the default binary installed by the conda package manager is the latest available, so if you need to get exactly the version corresponding to a specific robotology-superbuild distro release (for example for compatibility with an existing robot setup), please install the required versions by inspecting the version tables of the specific distro you are interested in https://icub-tech-iit.github.io/documentation/sw_versioning_table/ .
If you need to use the robotology packages with dependencies provided by other package managers, for example with the aptpackages on Debina/Ubuntu distributions, please install robotology-superbuild from source code as explained in the approprate section, as we do not provide binary packages for all software contained in the robotology-superbuild for the apt package manager.
We also support a deprecated way of installing binary packages just on Windows using dependencies provided by vcpkg, documentation for it can be found in doc/deprecated-installation-methods.md. However for new deployments we recommend to use conda binary packages also on Windows.
The first step to install robotology-superbuild from source is to download the robotology-superbuild code itself, and this is done through Git.
Once you install Git, you need to set your name and email to sign your commits, as this is required by the superbuild:
git config --global user.name FirstName LastName
git config --global user.email user@email.domain
Once git is configured, you can open a command line terminal. If you want to use the robotology-superbuild in rolling update mode, just clone the superbuild:
git clone https://github.com/robotology/robotology-superbuild
this will clone the superbuild in its default branch.
You can download and use the robotology-superbuild anywhere on your system, but if you are installing it
on an iCub robot laptop following the official iCub instructions, you should clone it in the /usr/local/src/robot directory.
If instead you want to use a specific release of the robotology superbuild, after you clone switch to use to a specific release tag:
git checkout v<YYYY.MM>
For the list of actually available tags, see the GitHub's releases page.
Once you cloned the repo, to go forward you can follow the different instructions on how to install robotology-superbuild from the source code, depending on your operating system and the package manager you want to use to install the required dependencies:
- Linux with dependencies provided by apt: use the superbuild on Debian/Ubuntu distributions installing the dependencies with apt,
- Linux, macOS or Windows with dependencies provided by conda-forge: use the superbuild on any supported operating system, installing the dependencies with conda or pixi package manager,
- Windows Subsystem For Linux: use the superbuild on Windows Subsystem For Linux.
The exact versions of the operating systems supported by the robotology-superbuild follow the one supported by the YARP library, that are documented in https://github.com/robotology/yarp/blob/master/.github/CONTRIBUTING.md#supported-systems .
Complete documentation on how to use a YCM-based superbuild is available in the YCM documentation.
When compiled from source, robotology-superbuild will download and build a number of software.
For each project, the repository will be downloaded in the src/<package_name> subdirectory of the superbuild root.
The build directory for a given project will be instead the src/<package_name> subdirectory of the superbuild build directory.
All the software packages are installed using the install directory of the build as installation prefix.
We also support an additional deprecated way of compiling the superbuild, on Windows using dependencies provided by vcpkg. Documentation for them can be found in doc/deprecated-installation-methods.md.
The following apt-based distributions are supported and tested by the robotology-superbuild:
- Ubuntu 20.04 (Focal Fossa)
- Ubuntu 22.04 (Jammy Jellyfish)
- Ubuntu 24.04 (Noble Numbat)
Other versions may be working, but they are not checked.
On Debian based systems (as Ubuntu) you can install the C++ toolchain, Git, CMake and Eigen (and other dependencies necessary for the software include in robotology-superbuild) using apt-get. This can be done by installing the packages listed in the apt.txt file using the following script:
cd robotology-superbuild
sudo bash ./scripts/install_apt_dependencies.sh
Besides the packages listed in apt.txt file, the script install_apt_dependencies.sh also installs some other packages depending on the distribution used, please inspect the script for more information.
For what regards CMake, the robotology-superbuild requires CMake 3.19 . If you are using a recent Debian-based system such as Ubuntu 22.04, the default CMake is recent enough and you do not need to do further steps.
If instead you use an older distro in which the default version of CMake is older, you can easily install a newer CMake version in several ways. For the following distributions, we recommend the following methods:
- Ubuntu 20.04 "Focal" : install a recent CMake via Kitware APT Repository, see https://apt.kitware.com/ .
For some profile or dependency specific CMake option you may need to install additional system dependencies, following the dependency-specific documentation listed in the following. If you do not want to enable an option, you should ignore the corresponding section and continue with the installation process.
Note that the ROBOTOLOGY_USES_GAZEBO option is enabled by default (except on Ubuntu 24.04 when installing with apt dependencies), so you should install Gazebo Classic unless you plan to disable this option.
On Linux with apt dependencies install Gazebo Classic, if you are on:
- Ubuntu 20.04
follow the instructions available at https://gazebosim.org/tutorials?tut=install_ubuntu . Make sure to install also the development files, i.e. libgazebo*-dev on Debian/Ubuntu.
Otherwise, if you are on other supported Debian/Ubuntu systems, just install the system provided gazebo package with:
sudo apt install libgazebo-dev
If you are on Ubuntu 24.04, please use conda if you want to install Gazebo Classic, as no Gazebo Classic packages are available via apt.
To install Modern Gazebo (gz-sim) on Ubuntu Jammy (22.04) and Noble (24.04) and other supported Debian/Ubuntu systems, follow the instructions available at https://gazebosim.org/docs/harmonic/install_ubuntu#binary-installation-on-ubuntu . Furthermore, you also need to install the cli11 dependency with:
sudo apt-get install libcli11-dev
Install Python and the necessary development files using the following command:
cd robotology-superbuild
sudo bash ./scripts/install_apt_python_dependencies.sh
On any Debian or Ubuntu based system, install PCL and VTK via
sudo apt install libpcl-dev
Install octave and the necessary development files using the following command:
sudo apt-get install octave-dev
Finally it is possible to install robotology software using the YCM superbuild:
cd robotology-superbuild
mkdir build
cd build
ccmake ../
source ./install/share/robotology-superbuild.sh
makeYou can configure the ccmake environment if you know you will use some particular set of software (put them in "ON"). See Superbuild CMake options for a list of available options.
The superbuild provides an automatically generated setup.sh sh script that will set all the necessary enviromental variables to use the software installed in the robotology-superbuild. To do so automatically for any new terminal that you open, append the following line to the .bashrc file:
source <directory-where-you-downloaded-robotology-superbuild>/build/install/share/robotology-superbuild/setup.sh
To use the updated .bashrc in your terminal you should run the following command:
user@host:~$ source ~/.bashrcIf may also be necessary to updates the cache of the dynamic linker:
user@host:~$ sudo ldconfigIf for any reason you do not want to use the provided setup.sh script and you want to manage your enviroment variables manually, please refer to the documentation available at doc/environment-variables-configuration.md .
If you want to use the conda package manager refer to doc/conda-forge.md document for instruction on how to compile the superbuild from source using the conda-forge provided dependencies, in particular the Source Installation section.
Otherwise, if you want to use the pixi package manager to install and build the superbuild, refer to the doc/pixi.md document.
The Windows Subsystem for Linux (wsl) lets developers run a GNU/Linux environment -- including most command-line tools, utilities, and applications -- directly on Windows, unmodified.
As all the software running on Linux distributions can run unmodified on Windows via WSL, to install the robotology-superbuild in WSL you can just install a Debian-based distribution for WSL, and then follow the instructions on how to install the robotology-superbuild on Linux, with dependencies provided either by apt or by conda. As the WSL enviroment is nevertheless different, there are a few things you need to care before using the robotology-superbuild on WSL.
Note that in the following we assume that you have configure your WSL to run graphical applications, as documented in https://learn.microsoft.com/en-us/windows/wsl/tutorials/gui-apps .
By default, the PATH enviroment variable in WSL will contain the path of the host Windows system, see microsoft/WSL#1640 and microsoft/WSL#1493. This can create problems,
as the CMake in WSL may find (incompatible) Windows CMake packages and try to use them, creating errors due to the compilation.
To avoid that, you can create in your WSL2 instance the /etc/wsl.conf file, and then populate it with the following content:
[interop]
appendWindowsPath = false
Note that you will need to restart your machine to make sure that this setting is taked into account.
If you want your YARP applications on WSL to connect to a yarpserver that you launched on the Windows native host (so on Command Prompt, not on WSL), you need to add the following line to your WSL's ~/.bashrc:
export WINDOWS_HOST=$(grep nameserver /etc/resolv.conf | awk '{print $2}')
yarp conf ${WINDOWS_HOST} 10000 > /dev/null 2>&1
Important: do not use this line if you are launching yarpserver directly on WSL.
If you are using the robotology-superbuild in its default branch and not from a release tag (i.e. in rolling update mode), to update the superbuild you need to first update the
robotology-superbuild repository itself with the git command:
git pull
After that, you will need to also run the equivalent of git pull on all the repositories managed by
the robotology-superbuild, you have to run in your build system the appropriate target.
To do this, make sure to be in the build directory of the robotology-superbuild and run:
make update-all
make
using make on Linux or macOS or
cmake --build . --target ALL_UPDATE
cmake --build .
using Visual Studio on Windows or
cmake --build . --target ALL_UPDATE
cmake --build .
using Xcode on macOS.
Note that the update will try to update all the software in the robotology-superbuild, and it will complain if the repository is not in the expected branch.
For this reason, if you are activly developing on a repository managed by the robotology-superbuild, remember to switch the YCM_EP_DEVEL_MODE_<package_name>
option to TRUE. This option will ensure that the superbuild will not try to automatically update the <package_name> repository. See https://robotology.github.io/ycm/gh-pages/git-master/manual/ycm-superbuild.7.html#developer-mode
for more details on this options.
Important
Before August 2024 the robotology-superbuild raised an error if you called make update-all or ninja update-all and there was repo with local modification with YCM_EP_DEVEL_MODE_<package> set to OFF. Since August 2024, instead the robotology-superbuild will silently discard the local modifications. To avoid losing data, never call the update-all target if you have local modifications in a package and you did not set YCM_EP_DEVEL_MODE_<package> to ON for that package.
By default, the robotology-superbuild uses the latest "stable" branches of the robotology repositories, but in some cases it may be necessary to use the "unstable" active development branches, or use some fixed tags. For this advanced functionalities, please refer to the documentation on changing the default project tags, available at doc/change-project-tags.md.
FAQs for robotology-superbuild can be found at doc/faqs.md.
For questions related to how to modify the rootology-superbuild itself, such as how to add a new package, how to do a release, check the Developers' FAQs document at doc/developers-faqs.md.
As robotology-superbuild is based on YCM, also YCM's FAQs could be relevant.
- Silvio Traversaro @traversaro