Oliver D. Reithmaier

New blogpost up, inspired by @eliocamp on how to make your #rstats projects fully computationally reproducible with docker, pak and renv.

Share, comment, feedback at will :)

https://k4tana.github.io/blog/2026/02/25/Reproducible-Docker-R.html

#statistics #usec #research #academicchatter

Fully Computationally Reproducible R Workflows with Docker, pak and renv (2026 Edition)

Hey Friend,

K4tana - Cutting Edge Research Blogs
Feb 25 at 1:50pmWeb
Show threadJordi RosellFeb 25
@odr_k4tana @eliocamp I did something similar in the past. One thing I miss and I have in my TODO... Is having my own binary repo and not having to build from source everytime.
Show threadOliver D. ReithmaierFeb 25
@jrosell @eliocamp yeah that's the biggest nuisance tbh.
Show threadElio CampitelliFeb 25
@jrosell @odr_k4tana You can always use posit public package manager, which provides binaries for Linux.
Show threadOliver D. ReithmaierFeb 25
@eliocamp @jrosell yes, but apparently (iirc) they only feature latest package versions. Once you go back (for whatever reason), you need to install from source
Show threadElio CampitelliFeb 25
@odr_k4tana @jrosell Ah, that sucks.
Show threadJordi RosellFeb 26
@eliocamp @odr_k4tana my take was to be able to save the used versions and have a private repo of binaries for all the projects.
Show threadElio CampitelliFeb 26
@jrosell @odr_k4tana Would using a private repo play well with renv and reproducibility? Wont renv::restore() fail without access to the repo?
Show threadElio CampitelliFeb 25
@odr_k4tana Neat! I'll need to review these steps to update reproducibility.rocks. lately I'm working 100% on hpc clusters so I can't really use docker easily.