Skip to content

Latest commit

 

History

History
298 lines (199 loc) · 10.3 KB

CompilationUbuntu.md

File metadata and controls

298 lines (199 loc) · 10.3 KB
Tutorials Home Previous Next

Compiling and installing libpointmatcher on Ubuntu

In short...

If you are used to development projects, here is what you need:

Name Version
(Tested Feb. 20, 2015)
Version
(Tested Sept. 6, 2016)
Version
(Tested Jan. 8, 2019)
Ubuntu 12.04.5 LTS (64 bit) 14.04.5 LTS (64 bit) 18.04.1 LTS (64 bit)
gcc 4.6.3 4.8.4 7.3.0
git 1.7.9.5 1.9.1 2.17.1
cmake 2.8.11.2 2.8.12.2 3.10.2
doxygen (opt.) 1.7.6.1 1.8.6-2 1.8.13-10
Dependency:
boost 1.48.0.2 1.54.0 1.65.1
eigen 3.0.5 3.2.0-8 3.3.4-4
libnabo from source from source from source

Note: we only support 64-bit systems because of some issues with Eigen. Other versions will most probably work but you'll have to try yourself to know for sure.

The rest of this tutorial will guide you through the different requirements step by step.

Detailed Installation Instructions

Some Basic Requirements

a. Installing Boost

Boost is a widely-used C++ library and is included in most Linux distributions. You can quickly check if Boost is installed on your system by running

ldconfig -p | grep libboost

If you see a list of results then you can skip to the next section. If not, you most likely have to install Boost.

Instructions for downloading and installing boost to Unix systems can be found here. Boost can also be installed as a package by running

sudo apt-get install libboost-all-dev

b. Installing Git

Git is a version control system similar to SVN designed for collaboration on large code projects. Because libpointmatcher is hosted on Github, you should the git application to keep track of code revisions, and bug fixes pushed to the public repository.

Git should already be installed on your system but you can check if it is by running

git --version 

If Git is installed, you should see a message of the form

git version 2.17.1

If not refer to the Git homepage for installation instructions or install via the package manager by running

sudo apt-get install git-core

c. Installing CMake

CMake is a cross-platform build system and is used for building the libpointmatcher library. Refer to the homepage for installation instructions, or you can once again use the package manager

sudo apt-get install cmake cmake-gui

NOTE: CMake has a GUI called cmake-gui which can be useful for configuring builds. We recommend you install this as well since it will be referred to in this tutorial.

NOTE 2: Some functionalities like find_package() (only useful if you intent to link with your own project) will only work if you have CMake 2.8.11 or over. On Ubuntu 12.04, only 2.8.7 is available using apt-get. If you intent to use such functionality, you will have to compile CMake from sources.

1. Installing Eigen

The Eigen linear algebra library is required before installing libpointmatcher and can be found here. Either download and compile Eigen following instructions from the package website or simply install package via apt by running:

sudo apt-get install libeigen3-dev

2. Installing libnabo

libnabo is a library for performing fast nearest-neighbor searches in low-dimensional spaces. It can be found here. Clone the source repository into a local directory of your choice.

mkdir ~/Libraries/
cd ~/Libraries
git clone git://github.com/ethz-asl/libnabo.git
cd libnabo

Now you can compile and install libnabo by entering the following commands

SRC_DIR=$PWD
BUILD_DIR=${SRC_DIR}/build
mkdir -p ${BUILD_DIR} && cd ${BUILD_DIR}
cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo ${SRC_DIR}
make

This will compile libnabo into a /build directory.

To make sure that everything is working properly, you can run the unit tests:

make test

This will run multiple nearest-neighbor searches performances and may take some minutes.

Then, install the library on your system by running the following command :

sudo make install

Note: If Eigen or Boost are not in their regular system locations you will have to indicate their location by setting the corresponding CMake flags. Go here to see how it can be achieve.

3. Installing libpointmatcher

First, you need to clone the source repository into a local directory. As an example, we reuse the Libraries directory that was created to contain the libnabo sources.

cd ~/Libraries/
git clone git://github.com/ethz-asl/libpointmatcher.git
cd libpointmatcher

But, before compiling libpointmatcher, a /build directory must be created. Just like with libnabo, run the following commands :

SRC_DIR=${PWD}
BUILD_DIR=${SRC_DIR}/build
mkdir -p ${BUILD_DIR} && cd ${BUILD_DIR}
cmake -D CMAKE_BUILD_TYPE=RelWithDebInfo ${SRC_DIR}

Before moving on to the compilation and installation steps, here are some optional features that you can enable.

Compiling the documentation (optional)

Libpointmatcher is documented directly in the source-code using Doxygen. If Doxygen is installed on your system, an html version of the documentation will be compiled in /usr/local/share/doc/libpointmatcher/. To install Doxygen on Ubuntu, run:

sudo apt-get install doxygen

You will also need LaTeX for the equations rendering :

sudo apt-get install texlive-full

After you have installed Doxygen and LaTeX, you can enable the documentation by setting the CMake variable GENERATE_API_DOC to TRUE. This can be achieved through CMake-GUI or by the command line:

cmake -D GENERATE_API_DOC=TRUE ${SRC_DIR}

Compiling libpointmatcher will generate the documentation, which you can simply open the /usr/local/share/doc/libpointmatcher/api/html/index.html file to view the API documentation in a browser.

Compiling unit tests (optionnal)

If you want to verify that the version of libpointmatcher you have compiled is stable, you can enable them by setting the CMake variable BUILD_TESTS to TRUE. It can be done with CMake-GUI or via the command line:

cmake -D BUILD_TESTS=TRUE ${SRC_DIR}

Then, once the compilation process is completed, the unit tests can be run with the following command line:

utest/utest --path ${SRC_DIR}/examples/data/

Compilation

Now, to compile libpointmatcher into the /build directory, run the following command:

make -j N

Note: It is highly recommended to add the -j N optionnal argument to the make command in order to speed up the compilation process. Replace N by the number of parallel jobs you want to compile at the same time.

Installation

Finally, to install libpointmatcher on your system, run the following command:

sudo make install

4. Possible Caveats

If Eigen, libnabo, yaml-cpp, or GTest are not found during the installation, you will have to manually supply their installation locations by setting the CMake flags. You can do so using the CMake-GUI.

cd build
cmake-gui .

alt text

You can then set EIGEN_INCLUDE_DIR, NABO_INCLUDE_DIR, NABO_LIBRARY, yaml-cpp_INCLUDE_DIRS, yaml-cpp_LIBRARIES to point to your installation directories as shown in the screenshot above. Then, generate the make files by clicking generate and rerun the following inside /build:

make
sudo make install

Having problems?

Some dependencies changed and we don't keep track of all combinations possible. Before reporting a problem, make sure to include the versions you are using. You can run the bash script ./utest/listVersionsUbuntu.sh and copy-paste its output when reporting an issue on github. You may need to ensure that the file is executable:

chmod +x ./utest/listVersionsUbuntu.sh
./utest/listVersionsUbuntu.sh

Here are the list of useful commands used in the bash script:

Ubuntu version:

lsb_release -r

32-bit or 64-bit architecture:

getconf LONG_BIT

Compiler version:

gcc --version

Git version:

git --version

CMake:

cmake --version

Boost version:

dpkg -s libboost-dev | grep Version

Eigen3:

dpkg -s libeigen3-dev | grep Version

Doxygen:

dpkg -s doxygen | grep Version