Reproducible Research Compendia via R packages


Hi everyone,

I had a great time meeting the group yesterday, and the discussion was very stimulating and thought-provoking for me. It was great fun to visit such a vibrant research community. Thanks again for the chance to join you!

My slides are online here:, and there are many links out to more details on the topics mentioned.

My template repo for my research compendia is here: Please clone/download it and write your next paper with it! It’s still a work in progress, and I got several good ideas from talking with you yesterday about how to improve and simplify, so I will be working on that. Don’t hesitate to reply here or email me ( with any questions or if you need it to do something in particular for your research.




Hi Ben,

Thank you so much for taking the time to talk with us and these resources. I learned a ton. I’ve always noticed those little green labels of Github pages saying “passing”, etc., but never known how people got them there or exactly what they meant. What a great feeling seeing it flick to green. I’ll be getting my students to implement on-the-fly code testing in their workflows asap. Docker in the future (baby steps).

Cheers. Josh.


Thanks Josh!

There’s a nice recent post by my colleague Carl Boettiger on a minimalist research compendia that he uses with students: It’s just R, GitHub and Travis, stripped down to the simplest bits necessary. Looks like a great entry point for getting students familiar with reproducible research and a good current tool kit.

I’m keen to know how you go with your students, do post back with links to some of your student’s work, if possible!




I posted my slides to twitter also, and got some very interesting suggestions from Vicky Steeves at NYU.

In brief, she mentioned that GitLab (a service like GitHub) has a built in continuous integration service that can use Docker. So that removes the need for travis-ci and circle-ci, that I described in my talk, potentially saving some time. Good examples here: &

The GitLab services look good, and I plan to experiment with it, but the R community is mostly on GitHub, so I think I’ll probably stay where the community is for the moment. Plus my travis-ci and circle-ci tooling is very efficient now, so I’m not sure that there’s a lot of time to be saved by switching.


Twitter thread with Vicky is here: (I can only post two URLs at a time)