Report from the Reproducible Builds Summit

For the Guix project we have just published a report from the Reproducible Builds Summit that took place ten days ago. (A more comprehensive report should eventually be available at .)

Reproducible builds are often regarded as a technical issue not necessarily relevant to the broader topic of reproducible scientific workflows. I believe this is misguided; on the contrary, there’s a continuum: reproducible science needs (among other things!) reproducible software deployment, and reproducible deployment needs reproducible builds — they are what makes it possible to independently verify that the software really corresponds to the expected source code.

The related issue of bootstrapping also ties in with research issues in programming languages, notably.

Feedback welcome!



From my point of view, reproducible computations in science and reproducible builds are exactly the same problem in different but overlapping application domains.