Continuous Integration with Jenkins
The software stack has grown to a size where it make sense to somehow track the state after each commit. The Jenkins Software allows us to do this. The installation can be found at here and all of Osmocom is tracked there. The goal is to have all projects/repositories of the osmocom umbrella tracked there. We are happy if people donate CPU time to be used as additional nodes to execute the compile jobs.
We are using Jenkins as it comes from the FreeBSD port system. It runs using the standalone servlet container (Winstone Servlet Engine) and is behind one (IPv6) or two (IPv4) cherokee reverse proxies. Currently there are two nodes FreeBSD8.2/AMD64 and Debian Squeeze/i386 ( running on the FreeBSD kernel, any weirdness might be a broken syscall implementation).
The osmo-deps.sh script was created to clone/update a dependency in the workspace of a build node. The script will try to git-clone, git-fetch, git-reset --hard the dependency.
if ! test -d $1; then git clone git://git.osmocom.org/$1 $1 fi cd $1 git fetch origin git reset --hard origin/master
FreeBSD 8.2 and Osmocom
Not everything is required to build on !GNU/Linux systems, for SIMtrace a fake libusb-1.0.pc was created to point to the /usr/include,/usr/lib for libusb, all dependencies (fftw, erlang, autotools...) were installed through the port system
Normal Debian/Squeeze? installation, all dependencies installed via the apt/dpkg package system. The arm-elf-gcc toolchain is installed using the toolchain script of the bb.osmocom.org wiki.