Analysis reproducibility with containers
Discussion started during the zoom call. To be continued here.
@stefano: shall we archive the containers recipe or the full images if we want to ensure reproducibility through time?
Stefano: IMHO the source of truth should be the recipes/Dockerfiles, and images should be archived as a "quick access" for the build artifact..
Kay: to combine both of your remarks, I propose to store both recipes and images (first for maintenance, second to run) as als @stefano confirms
Stefano: it depends what you are doing with your software. If you are providing a service, then definitely it is a security issue. But if you are running a scientific analysis instead, updating versions is often counter-productive and undermines reproducibility of results or event the entire validity of work..@Christian it depends what you are doing with your software. If you are providing a service, then definitely it is a security issue. But if you are running a scientific analysis instead, updating versions is often counter-productive and undermines reproducibility of results or event the entire validity of work..