Hmm... That looks like a big project. For something like that, I would start with a pilot "application" (maybe a demo or a tiny simulation or some such) to get started. And, once that was working well, using that experience to tackle the larger project. (And, finally, ensuring that the larger project is complete by finding ways and/or people to adequately exercise each of the implementations.)
(That said, personally, my personal interest would be more focused on https://gmplib.org/ ...) Anyways, I guess my point is that you'd need to be highly motivated to pull off something like this. Still, if you want to go there, and get stuck, I imagine many of us would be happy to try and help you work your way through technical obstacles. For starters, you might want to look at stuff like https://www.jsoftware.com/help/user/call_procedure.htm or https://code.jsoftware.com/wiki/Guides/DLLs/Calling_DLLs or https://www.jsoftware.com/help/jforc/calling_external_programs.htm or https://www.jsoftware.com/help/user/lib_dll.htm There's some version dependencies, and sometimes you run into something which needs attention. (Which brings up another issue -- it's probably better to start focusing on one specific operating system, and its conventions, and then once that's working making another copy and doing that for another operating system, and only then, or maybe after a third port, working to combine the copies. The combining part is much easier to do when you're merging working code.) If you already knew all of this, please accept my apologies. Good luck, -- Raul On Wed, Feb 24, 2021 at 8:07 AM Emir U <e...@usgroupltd.uk> wrote: > > Hi fellow statisticians and ML folk , I wondered if there may be appetite to > co-develop J bindings for GNU Scientific? Its a 1000 plus functions which > covers practically everything: distributions, stats, regression, > minimisation, non-linear fitting, LA, approximations, signal proc, lots else. > Full list of functions here: > > https://www.gnu.org/software/gsl/doc/html/index.html > > I'm a neophyte but I'm happy to pitch in if someone experienced would set the > template and gatekeep the project. I think it'd set J up a serious language > for statistics. It'd combine a world class library with unparalleled > conciseness and productivity. Is there any experienced interest in this > direction, or an alternative direction? > > Emir > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
