Simon Tournier

GNU Guix is our beloved reproducible software deployment toolbox. It is a package manager also able to create isolated computational environments. It is even able to produce pack containers. Or, it can be a complete system. In addition of focusing on reproducibility, it provides some key features as roll-back, declarative configuration, or time-machine, to name some. GNU Guix respects the four essential freedoms: run the program as you wish; study how the program works and change it for computing as you wish; share copies for helping others; distribute modified versions for improving.

Open science considers that scientific practices must be more open, transparent, collaborative and inclusive, coupled with more accessible and verifiable scientific knowledge subject to scrutiny and critique. Such enterprise will improve the quality, reproducibility and impact of science, and thereby the reliability of the evidence needed for robust decision-making and policy and increased trust in science. Quoting UNESCO.

Wait, Guix and Open science seems so well matched. This presentation is an attempt to illustrate how the Guix project is helping Open science. We emphasize two features of Guix that implement Open science principles: transparent and long-term.

This presentation illustrates how the Guix project provides software deployment for reproducible research. GNU Guix takes care about the computational environment from the package management to producing container (Docker, Singularity) and also being an alternative for virtual environment.

Open science means transparent and collective; transparent because a scientific result needs to be scrutinized and studied bug-to-bug, and collective because an independent observer must observe the same result–at least when speaking about computational processing–and this observation needs to be sustainable. Guix is an attempt to implement, for the computational environment, these two items.

By applying functional programming concepts to package management, pioneered by Nix, Guix is deeply transparent by design. The whole computational environment is captured and its inspection, from source code to binary, becomes tractable. As a nice consequence, this computational environment is reproducible from one machine to another. What about missing source code? Guix is able to transparently fallback to the Software Heritage archive. The reproduction of this computational environment becomes sustainable–to some extent.

In short, this presentation try to convince how Guix and 3 command lines can helps open science.