The GitHub repository of JSBSim is reachable at this link: github.com/JSBSim-Team/jsbsim.
You need to have the Git software installed. Git is a version control software, a system that records changes to a file or set of files over time so that you can recall specific versions later. The JSBSim software source code files are being version controlled by Git.
To install Git go to its download site and grab the version for your platform. You can choose to use Git locally on your computer in two ways: via one of the GUI clients, or through a command shell (e.g. a Bash shell on Linux or Windows).
Once you have installed Git, assuming you are going to use Git from the command shell, the JSBSim source code public repository can be cloned from GitHub
The Git command to clone the repo is (HTTPS mode)
> git clone https://github.com/JSBSim-Team/jsbsim.git jsbsim-code
or (SSH mode)
> git clone [email protected]:JSBSim-Team/jsbsim.git jsbsim-code
JSBSim can either be built with CMake or Microsoft Visual Studio. If you are using a Mac OSX or a Linux platform, you must use CMake. If you are a Windows user you can use either one.
JSBSim is coded in standard C++11/C99 and has no dependencies, so all you need is a C/C++ compiler installed on your platform.
CMake is a multiplatform tool to build and test software. It can produce files to build JSBSim with GNU make or Microsoft Visual Studio. To keep the build files separated from the source code, it is preferable to build JSBSim in a separate directory.
> cd jsbsim-code
> mkdir build
> cd build
CMake does not build software, it produces files for a multitude of build tools. The following commands are assuming that you are using GNU make to build JSBSim.
First, you should invoke CMake and then execute make
> cmake ..
> make
This will compile the various classes, and build the JSBSim application which will be located in build/src
CMake can use a number of parameters to tune the build of JSBSim. Different options are presented below. You can use them independently or any combination thereof depending on your needs.
If you want to set compiler options, you can pass flags to CMake to build a Debug
version of JSBSim. JSBSim also uses C for some code, you can set options for both the C++ and the C compiler.
> cmake -DCMAKE_CXX_FLAGS_DEBUG="-g -Wall" -DCMAKE_C_FLAGS_DEBUG="-g -Wall" -DCMAKE_BUILD_TYPE=Debug ..
> make
Or alternatively you can build a Release
version of JSBSim and request GNU Make to use 4 cores to build the executable faster.
> cmake -DCMAKE_CXX_FLAGS_RELEASE="-O3 -march=native -mtune=native" -DCMAKE_C_FLAGS_RELEASE="-O3 -march=native -mtune=native" -DCMAKE_BUILD_TYPE=Release ..
> make -j4
JSBSim uses the Expat library to read XML files. The Expat source code is provided with JSBSim source code and is compiled along with JSBSim during its build. However, if Expat is already installed on your platform you might prefer to use your system Expat library in order to avoid duplication. In that case, you should pass the SYSTEM_EXPAT
flag to CMake:
> cmake -DSYSTEM_EXPAT=ON ..
> make
Most of JSBSim code can be built as a shared library, so that the executable JSBSim
and the Python module can share the same library which reduce the memory and disk space consumption.
The option BUILD_SHARED_LIBS
must then be passed to CMake
> cmake -DBUILD_SHARED_LIBS=ON ..
> make
A Python module of JSBSim can also be built by CMake. For that, you need Cython installed on your platform. CMake will automatically detect Cython and build the Python module.
From Visual Studio, you can open the project file JSBSim.vcxproj
to open a project for JSBSim. The project file will setup Visual Studio for building the JSBSim executable.
Note 1: JSBSim official build tool is CMake. Visual Studio project files are provided as a convenience and are not guaranteed to be up to date with the code.
Note 2: Since Visual Studio 2017, Microsoft has included CMake so you can build JSBSim on VS2017 directly from the CMake file.
For more detailed instructions on using Visual Studio project files and CMake via Visual Studio to build JSBSim take a look at the following documentation link - Building using Visual Studio.
JSBSim comes with a test suite to automatically check that the build is correct. This test suite is located in the tests
directory and is coded in Python so you need to build the Python module of JSBSim first.
The tests are also using numpy
, pandas
and scipy
so you need these Python modules to be installed on your system.
Make sure that the JSBSim shared library is in the path of the dynamic linker before running the tests (see the procedure to add the JSBSim library to the dynamic linker path if you get multiple failures of the tests).
The test suite can then be run using ctest
in the build
directory. Tests can also be run in parallel on several cores (4 in the example below) using the option -j
> ctest -j4
Tests ran with ctest
include C++ unit tests that are written with the CxxTest test framework.
A code coverage report is automatically generated and is available at: https://jsbsim-team.github.io/jsbsim/coverage/
Once JSBSim is built and tested, you can install the C++ headers and library. For that, you can invoke GNU make from the build
directory
> make install
By default, CMake copies the files to a location where the headers and library are available platform wide (typically /usr/include
, /usr/lib
or /usr/local/include
, /usr/local/lib
for *nix OSes). If you want to install the files in another location you can pass the flag CMAKE_INSTALL_PREFIX
to cmake.
cmake -DCMAKE_INSTALL_PREFIX=/path/to/install ..
make
make install
If you plan to install the Python module of JSBSim in addition to the C++ headers and library, then you must pass the flag INSTALL_JSBSIM_PYTHON_MODULE
to CMake. This is the procedure you should follow if you plan to package JSBSim with CPack.
> cmake -DINSTALL_JSBSIM_PYTHON_MODULE=ON ..
> make
> make install
Note: make install
will attempt to override Python virtual environments in order to install the Python module platform wide (i.e. in a directory such as /usr/lib/python
). If you want the Python module installation process to comply with your virtual environment, you should use the Python distutils
as described below.
Alternatively, the Python module can be installed manually by invoking the Python distutils
from the build
directory. The installation will ignore the option CMAKE_INSTALL_PREFIX
and will be performed independently of the setting of INSTALL_JSBSIM_PYTHON_MODULE
.
This is the procedure you should follow if you are using Python virtual environments.
> python python/setup.py install --skip-build
The option --skip-build
avoids to rebuild the module.
JSBSim can also be packaged for releases. This is done automatically by Travis CI for the Ubuntu 14.04 LTS platform and the resulting Debian packages are available for download on the JSBSim GitHub project page.
The packaging can be done by passing the option CPACK_GENERATOR
to CMake then invoking CPack.
At the moment, only RPM and Debian packages are supported by JSBSim.
> cmake -DCPACK_GENERATOR=DEB .. # or RPM
> make
> cpack
The following packages are then built (with the extension .rpm
if you selected the RPM generator)
JSBSim_[version].[platform].[architecture].deb
which contains the executablesJSBSim
andaeromatic
(and shared libraries ifBUILD_SHARED_LIBS
was set toON
)JSBSim-devel_[version].[platform].[architecture].deb
which contains the files for C++ development headers (and the static library ifBUILD_SHARED_LIBS
was not set toON
)python[2-3]-JSBSim_[version].[platform].[architecture].deb
which contains the JSBSim Python module ifINSTALL_JSBSIM_PYTHON_MODULE
was set toON
The JSBSim C++ API documentation is built from the source code with Doxygen and is automatically published on GitHub each time a commit is pushed to the JSBSim GitHub project. It can be viewed here:
https://jsbsim-team.github.io/jsbsim/.
If you modify the documentation, you might need to generate the documentation locally in which case you should run the following command after cmake
has been executed
> make doc
The HTML documentation will then be available in the directory build/documentation/html
. Note that you need Doxygen and Graphviz to be installed.
Q: I ran ctest
from my build directory and it reports one or several tests failures. The problem is that ctest
does not seem to give any details about the reason why the tests failed.
A: All the output issued by tests run by ctest
are logged in the file Testing/Temporary/LastTestsFailed.log
.
Q: One of the tests ran by ctest
is failing and I am trying to debug it. But ctest
always executes all the tests and it takes much time. To iterate faster, I would like to focus only on the test that fails. How do I tell ctest
to skip all the tests but the test that fails ?
A: You can filter the tests executed by ctest
with the options -R
, -E
and -I
. Say you want to run exclusively TestDensityAltitude.py
then you can use the following command
> ctest -R TestDensityAltitude
If you need to run all the tests which name contains Altitude
, you can use
> ctest -R Altitude
If you need to run all the tests but thoses which name contains Altitude
, you can use
> ctest -E Altitude
If you need to run tests #12 to #14, you can use
> ctest -I 12,14
You can find more informations about ctest
from its manual page
Q: Before running make install
, I want to execute ctest
but most of the tests fail. And when I check in the file Testing/Temporary/LastTestsFailed.log
it reports many Python ImportError
such as
31: Traceback (most recent call last):
31: File "TestTurbine.py", line 23, in <module>
31: from JSBSim_utils import JSBSimTestCase, CreateFDM, RunTest, append_xml
31: File "/xxx/build/tests/JSBSim_utils.py", line 24, in <module>
31: import jsbsim
31: ImportError: libJSBSim.so.1: cannot open shared object file: No such file or directory
A: This error means that the dynamic linker ld.so
cannot find the JSBSim shared library (libJSBSim.so.1
) so you must tell it where it is located. For that purpose, you can use the environment variable LD_LIBRARY_PATH
like below
> LD_LIBRARY_PATH=$PWD/src ctest
The command above will succeed if you execute it from your build directory in which case the environment variable PWD
will contain the path to your build directory.
To avoid prepending every ctest
command with LD_LIBRARY_PATH=$PWD/src
you can execute the command below to store the modification to the LD_LIBRARY_PATH
environment variable. This command must be executed from your build
directory in order to store the appropriate path.
> export LD_LIBRARY_PATH=$PWD/src:$LD_LIBRARY_PATH
Please note that as soon as your shell session will be terminated, the modification to LD_LIBRARY_PATH
will be lost.