mraa
1.8.0
Low Level Skeleton Library for Communication on GNU/Linux platforms
|
libmraa uses cmake in order to make compilation relatively painless. CMake runs build out of tree so the recommended way is to clone from git and make a build/
directory inside the clone directory.
For building imraa check buildingimraa page.
Not all these are required but if you're unsure of what you're doing this is what you'll need:
For Debian-like distros the below command installs the basic set:
```bash sudo apt-get install git build-essential swig3.0 python-dev nodejs-dev cmake libjson-c-dev ```
To build the documentation you'll also need:
If this goes wrong and you have all the dependencies installed, then please file an issue with the full output of cmake ..
and make
or however far you got.
After that you can install built files (into default path) by running:
```bash sudo make install ```
See flags for adjusting install paths in the section below.
Currently our install logic puts Python bindings into standard paths, which do not work on Debian due to their policy.
We are working on a permanent solution, in the meantime please use this command after make install
to link installed modules where Debian's Python expects them:
```bash sudo ln -s <your install prefix, e.g. /usr>/lib/python2.7/site-packages/* /usr/lib/python2.7/dist-packages ```
Same approach works for Python 3, you'll just need to adjust the version number in the path accordingly.
Our CMake configuration has a number of options, cmake-gui
or ccmake
(cmake -i
is no longer with us :() can show you all the options. A few of the more common ones are listed below. Note that when the option starts with CMAKE_
it's an option that is made available by CMake and will be similar in all CMake projects. You need to add them after cmake
but before ..
A few recommended options:
Changing install path from /usr/local
to /usr
: -DCMAKE_INSTALL_PREFIX:PATH=/usr
Building debug build - adds -g
and disables optimisations - this will force a full rebuild: -DCMAKE_BUILD_TYPE=DEBUG
Using clang
instead of gcc
: -DCMAKE_C_COMPILER=/usr/bin/clang -DCMAKE_CXX_COMPILER=/usr/bin/clang++
Building with an older version of SWIG (< 3.0.2) requires the disabling of JavaScript: -DBUILDSWIGNODE=OFF
Disabling Python module building: -DBUILDSWIGPYTHON=OFF
Building doc, this will require SPHINX & Doxygen: -DBUILDDOC=ON
You will also require clone git submodules from your existing checkout: git submodule update --init --recursive
Then from doxygen2jsdoc dir: npm install
Then from doxyport dir: make setup
Override build architecture (this is useful because on x86 ARM code is not compiled so use this flag to force the target arch) -DBUILDARCH=arm
You can also enable -Wall for gcc before running cmake by exporting your wanted CC flags to the CC env var export CC="gcc -Wall"
Sometimes it's nice to build a static libary, on Linux systems just set -DBUILD_SHARED_LIBS=OFF
You'll need at least SWIG version 3.0.2 and we recommend 3.0.5 to build the JavaScript & Python modules. If your version of SWIG is older than this then please see above for disabling SWIGNODE
. Otherwise you will get a weird build failure when building the JavaScript module. The Python module builds with SWIG 2.x but we don't test it.
During the build, we'll assume you're building from git, note that if you compile with git
installed your version of mraa will be versioned with git desribe --tag
to make it easy for intentification. You can easily modify version.c in build/src. If you don't build from a git tree then you will simply have a version which matches the latest released version of mraa.
In order to compile with a Yocto/OE toolchain use the following toolchain file. This works well on the Edison 1.7.2 SDK. First source the environment file, then use our CMake toolchain file.
This is the procedure to submit a build to Coverity. You'll need to install coverity-submit
for your OS.
Have JAVA_HOME set to JDK install directory. Most distributions set this from /etc/profile.d/
and have a way of switching between alternatives. We support both OpenJDK and Oracle's JDK. On Arch Linux with OpenJDK 8 you'll have to set this yourself like this:
Then use the CMake configuration flag: -DBUILDSWIGJAVA=ON
To compile Example.java
To run, make sure libmraajava.so
is in LD_LIBRARY_PATH
If you want to add or improve Java bindings for mraa, please follow the Creating Java Bindings Guide.
cpack
You can get cpack
to generate an IPK or RPM package fairly easily if you have the correct packaging tools
To use RPM simply enable the RPM option. You'll need rpmbuild
installed on your build machine.
Requirements:
The Things Native Library contains a CMake find_package module FindAndroidThings.cmake. Make sure the directory containing this module is added to the CMAKE_MODULE_PATH.
You can use docker
and docker-compose
to generate a complete build environment for mraa without having to install any other tool.
Requirements:
NOTE: docker-compose is an optional requirement. It actually make running complex docker build and run command easier. But you can just use docker to build and run.
tl;dr: Just use this commands to build mraa:
```sh
$ docker-compose run doc
$ docker-compose run python2
$ docker-compose run python3
$ docker-compose run java
$ docker-compose run node4
$ docker-compose run node5
$ docker-compose run node6
$ docker-compose run android
$ docker-compose run sonar-scan ```
docker-compose will take a look at the docker-compose.yaml
file in the repository root directory, pull the required docker image, and run an specific command to build mraa for the requested target. Once the build is completed, you will have a build/
folder in the repository root with all the compiled code. This build/
folder is created by using a docker volume. The build\
folder contents is reused each time you execute docker-compose run [TARGET]
. To know more about volumes in Docker, visit the Docker Volume Documentation.
You can also start an interactive session inside the docker container if you need to run some custom build commands:
```sh
$ docker-compose run python2 bash
$ cd build && cmake -DBUILDSWIGPYTHON=ON .. && make clean all ```
If you don't want to use docker-compose, you can also use docker run
to build mraa. For example, to build mraa for python2, you can do:
```sh
$ docker run \ –volume=:/usr/src/app \ –env BUILDSWIG=ON \ –env BUILDSWIGPYTHON=ON \ –env BUILDSWIGJAVA=OFF \ –env BUILDSWIGNODE=OFF \ dnoliver/mraa-python \ bash -c "./scripts/run-cmake.sh && make -Cbuild _python2-mraa" ```
If, for some reason, you are behind a proxy, find below a list of common problems related to proxy settings:
docker cannot pull images from docker.io
Visit this link to configure docker daemon behind a proxy.
docker run fails to access the internet
docker-compose will automatically take http_proxy
, https_proxy
, and no_proxy
environment variables and use it as build arguments. Be sure to properly configure this variables before building.
docker, unlinke docker-compose, do not take the proxy settings from the environment automatically. You need to send them as environment arguments:
```sh
$ docker run \ –volume=:/usr/src/app \ –env BUILDSWIG=ON \ –env BUILDSWIGPYTHON=ON \ –env BUILDSWIGJAVA=OFF \ –env BUILDSWIGNODE=OFF \ –env http_proxy=$http_proxy \ –env https_proxy=$https_proxy \ –env no_proxy=$no_proxy \ dnoliver/mraa-python \ bash -c "./scripts/run-cmake.sh && make -Cbuild _python2-mraa" ```