Consider creating an R package ranking system

[bschilder]

I'm beginning to have real concerns over the usefulness and legitimacy of CRAN as an arbiter of package quality (see here), since they seem to have quite a few ongoing issues themselves (both technical and organisational). Also, because code coverage isn't considered as one of their criterion of acceptance, they are essentially penalizing individuals who have more comprehensive unit tests (especially in situations that the test fail due to issues with the CRAN servers, such as unstable internet connections, missing system deps, and inconsistent version pushing to each OS runner).

It makes me wonder if the R community would be better served by some sort of decentralised package checking and distribution service. For example, I could add a step to rworkflows that (if all checks pass) the package automatically gets listed on the "rworkflows approved" list of R packages. Rather than simply being pass/fail it would include additional markers of package quality and usability like code coverage, existence of container, passing X,Y,Z OS runners. This would allow for a publicly displayed ranking system where people can see how their packages stack up relative to the rest of the community. Badges could be issued for meeting certain criterion (gold, silver, bronze, etc) which would link to reports on how they can further improve the ranking of their packages (eg. increase code coverage, pass on Windows, pass CRAN checks).

[bschilder]

rdrr seems to have a way of getting mass info on R packages across official R package repos and GitHub. might be able to leverage this.
https://rdrr.io/