search

pveber / ocaml-r

Objective Caml binding embedding the R interpreter
GNU General Public License v3.0
35 stars 4 forks source link

I love you ! #17

Open gl-yziquel opened 12 months ago

gl-yziquel commented 12 months ago

Thank you so much for having kept this code alive !

pveber commented 12 months ago

Ha ha, well it was very useful to me more than once, so I should be the one to thank you :)

The API was rewritten several times, the current version is intended to be abstracted away from R, and friendlier when using from OCaml. Also, the R internals library deprecated quite a few types and functions, which forced me to remove most of the introspection code, as well as a mechanism to automate binding generation which I had written. But all in all, the main functionality (being able to call an R function from ocaml) is still working just fine.

gl-yziquel commented 12 months ago

Gee !! I just noted a few days ago that you used this code. I assumed that this code was more or less dead and that no one would bother looking into it.

Life has been rather unkind to me, so I'm only getting back into serious programming + stats now. Dunno if I'll reuse that code, though I hope I will. The ranger code sample using ocaml-R for random forests that has been mentioned in some comment would be an interesting toy project for me to get back into that code.

I wonder, though: I've had my doubts one the soundness of the interactions of the two GCs (and if I remember well, I did bang my head on trying to catch R exceptions on the OCaml side with the 2 GCs moving around). I guess there has meanwhile been improvements in OCaml's GC. I've heard it went concurrent since these prehistoric times 14 years ago. So: did you experience GC integration unsoundness with that code ? And did you get some Lwt async code running for days or months with that code (it was my original intent) ?

Anyhow, thanks accepted. Please accept mine.