I’ve written some Stan functions that implement various probability distributions which are not built into Stan. I’ve emailed them to my colleagues and they are using them in a variety of different models. However, it’s tricky to keep everyone up-to-date as I fix bugs and add new features. Therefore, I’ve created an R package that contains an R function for each density function that returns a string containing the Stan code. This is nice as I can also include the R version of each density function in the same package and easily test they produce the same results.
However, I feel like there must be a better way of distributing snippets of Stan code. Is there a better way of doing this or any existing examples I can look at for inspiration?
I have looked at rstanarm and brms, but they are doing something more complex than what I hope to achieve. I don’t want to create or distribute full models. I just want to make individual function definitions available to people.