rOpenSci review?

bayesplot
shinystan
loo

#1

rOpenSci is another NumFOCUS project that is offering to do “peer review” of R software for data science. Here’s an

I figure any feedback is good. We can always ignore it if it’s not useful. Seems what they talk about is going to be a very very light review compared to proper production code review you’d see at a company that cared about its software product.


#2

Hmm, I always thought of ROpenSci as a curator or like a “journal”. You bring them a package, they determine if it’s a good fit, then it gets a peer-review process and becomes an ROpenSci package. I’m not sure how that would interact with another organization… But maybe I misunderstand what they do.

I’ve reviewed for them. I’m not sure what a proper production review entails. I followed a checklist of typical concerns, read over and ran every line of code, suggested refactorings and improvements to unit tests and documentation, and wrote up my findings like an article review. It was actually pretty extensive. It might depend on the reviewer too though.


#3

Thanks for the insider perspective. I think that’s the drill as advertised. I think the “becomes an ROpenSci package” means they list as something they’ve “peer reviewed.” That we don’t really care so much about. On the other hand, free extensive external code audits always seem like a good idea. I wish someone did this for C++.


#4

Yeah I’d be intetested in this. An expert code review is always valuable in
my opinion. And the ROpenSci people are certainly R experts.