On Tue, 2017-05-02 at 10:07 +0000, Holger Levsen wrote: > On Tue, May 02, 2017 at 11:04:48AM +0100, Ghislain Vaillant wrote: > > In a nutshell, take a scientific experiment (data + processing > > pipeline), create a single archive out of it (packing step), and enable > > a different machine to reproduce the experiment (unpacking step). > > and how is it different from zip? (or tar|gzip) > > those can also be used to (un)pack reproducible archives…
The unpacking step includes fetching the necessary dependencies, and spawning a run of the experiment (locally, or on Vagrant / Docker) automatically. It's a step forward compared to providing a tar with code and data + custom deployment script, which researchers are usually quite terrible at writing. Ghis
