It can be a bit weird at first, so my hope with this document is to break down the git repository, and walk you through the steps in building a new image and making it accessible to the collaboration!
Go ahead and clone the repository anywhere you like:
$ git clone https://gitlab.com/supercdms/CompInfrastructure/cdms-jupyterlab.git
-
README, RELEASE, LICENSE - the usual stuff.
-
A build.sh script which does everything for us, just needs to be called from command line:
$ bash build.sh- The image version can be tweaked in here, it's the variable
$IMAGE_VER
-
The main course is the Dockerfile, which is a series of instructions for the Docker daemon when it builds an image.
- This is where our packages get installed
-
scripts/clone-repos.sh locally clones CMDS repositories from GitLab, to be copied into the image later
-
scripts/rootenv.sh is basically a boiled down version of ROOT's
thisroot.sh, which can be sourced from any Python and should allow importing ROOT -
hooks/post-hook.sh
- Creates symlinks from
/optto$HOME/Tutorialsso users have some reference material in their jupyter homespace- also makes sure this directory is read only, to prevent breaking edits to the global version
- Initializes the custom bash environment
- Creates symlinks from
-
hooks/launch.bash and kernels/py3-ROOT work together to configure the jupyter notebook kernel