After reading the article, I agree with Danielle and Claus

One thing that isn't mentioned is the different programming paradigms.

R seems to be built more toward functional programming, while python seems to be more toward OOP.

With that in mind, it doesn't bother me much about R OOP problems.

People often ask me if I will move rxode2 into python, and I say because of NSE it would be difficult. Julia has NSE, though so it is possible there I think.

I think it is really is the NSE, meaning the R code isn't actually evaluated and you can interpret it in any way you want.

In the case of rxode2 it is to use very simple syntax for odes

d/dt(central) = -kel*central

Thanks :)

Thanks! The static linking may alleviate the same problem, and I think CRAN would be in favor of it over ABI. However, if the binary gets too big, they may balk (just my opinion though).

In particular look at blog.nlmixr2.org/blog/2024-09...

Yes, I have, but there are many issues that may be created by this sort of linking. Here is a different solution: blog.nlmixr2.org/blog/2024-09...