General R interface?

[malcolmbarrett]

Hi @jolars

Exciting project (as are several of your other recent ones)! I am interested in seeing how it goes with a couple of R projects.

Are you intending to have a general R interface to basin? Or would you recommend calling basin directly?

[jolars]

Thanks! I have thought about this and the reason there isn't already an R package (or Python one for that matter) is really that some of the nice features about Basin disappear when you have to provide a stable runtime interface. Things like:

• choosing your linear algebra backend
• plugging in a solver in one of the memetic (injected) global solvers
• working with constraints
• adding an observer

So for me that limits the motivation for creating an R package for Basin. That being said, I guess the one thing that could still motivate this is if there are solvers in Basin that aren't available in other R packages (or are better implemented in Basin). Plus maybe if it would be nice to have another general interface (with drop-in alternatives to optim() and constrOptim())?

Do you have a use case in mind yourself? What would you hope to use Basin for in R (if it was available)?

For a package author who's willing to add some Rust code, I would definitely recommend using Basin directly.

[malcolmbarrett]

I ended up calling basic directly (here) in Rust. Thanks for the advice! I will close this for now.

[jolars]

Great! Just for the record, I am not ruling out adding an R package sometime in the future, I am just not sure right now that there would be an audience for it or that it would fill a gap in the current landscape.