Couple of weeks ago I wrote about all these new tools that showed up in the R ecosystem, that seemed to aim to fill the gap for modern tooling and better developer experience. Then I bumped in the excellent Python project template by Gemma Danks, and I thought to myself, why not make something similar for R? So here it is, an R project template with a focus on modularity and tooling.
But why I am presenting this with Homer Simpson’s spice rack? Well, it is a project template, and it works, but it seems many things can be improved. For example:
- Test coverage cannot be registered with codecov because it seems the R package covr does not work in this not-R-pacakage project structure. - pkgdown cannot be used for documentation for the same reason as above. - devtools::check() does not work because it expects a package structure with a DESCRIPTION file and an R/ directory for source code, which this template does not have.
Workarounds are of course possible, but these are kind of a goto tools for R projects, and I think many people would expect them to just work.
Even with these limitations, I think there is the learning curve for many to use box for writing modular R code, with box itself last time released in 2024.
But it works, and it is fun to see that it works. And it may be an alternative path for writing R code when we would like to avoid the classic package structure.
Code is here: R project template.
Let me know if you try it or just contribute on the repository.