Clone repositories in initializer script and do not preclone them

[ekiwi]

Currently the vm is shipped with our repositories already on the disk.

This leads to problems, when things in our repository structure change as is often the case in the beginning of a new season. E.g. this season we replaced the season2016 repo with a different one.
Now everyone using the new VM needs to delete the season2016 repo and run gits populate to get the latest code. A gits pull is not enough.

Thus I'd like to propose that we change the initializer script to clone the repositories when it is run instead of just updating them. As this normally happens in our room with a 1GBit/s connection to our server, this should not take much longer than just syncing the latest changes but is less error prone as pointed out above.

[salkinium]

The current solution bugs me as well, but gits populate is being called in init.sh.

The very first VM actually first downloaded the initializer script from the server, which was then executed.
It was a super secure download over only http ;-)

Perhaps you can add a new initializer script to our software.git umbrella repo, and git pull this, before executing the new script.
This allows "updating" of an already built VM.