Just starting dipping my toes into the lispy languages pool (aside from elisp), but my understanding all the advatanges for using LLM coding with Clojure would be just as, if not even more(?), applicable to CL or Racket?
But to answer your question, I think the thesis OP is basing his argument on is that Clojure is a very mutation-averse language and (unlike CL) it doesn’t really do any overly fancy control flow flips (e.g. conditions). This combined with the fact that most data structures use the exact same setters and getters lets the LLM reason better.
(Mind you, I can’t really vouch for the correctness of this assessment, I’ve never used any Lisps with LLMs, just how I understood the article.)
Just starting dipping my toes into the lispy languages pool (aside from elisp), but my understanding all the advatanges for using LLM coding with Clojure would be just as, if not even more(?), applicable to CL or Racket?
score 1 · canonical
There was indeed a post about that recently.
But to answer your question, I think the thesis OP is basing his argument on is that Clojure is a very mutation-averse language and (unlike CL) it doesn’t really do any overly fancy control flow flips (e.g. conditions). This combined with the fact that most data structures use the exact same setters and getters lets the LLM reason better.
(Mind you, I can’t really vouch for the correctness of this assessment, I’ve never used any Lisps with LLMs, just how I understood the article.)
score 1 · canonical
Thanks, a bit over my head, but interesting :)
score 1 · canonical