Hacker News new | comments | show | ask | jobs | submit login

> ...producing the same result as the Clojure's seq-based fns. The difference is that, reduce being eager, and these reducers fns being out of the seq game, there's no per-step allocation overhead, so it's faster. Laziness is great when you need it, but when you don't you shouldn't have to pay for it.

Looking forward to trying this out. I've been implementing another language in Clojure (tho just experimenting for the moment). It's a non-lazy language (Groovy) so I have reduce all over the place, e.g.

  (defn multiply [a b]
    (cond (and (number? a) (number? b)) (* a b)
          (and (vector? a) (integer? b)) (reduce conj [] (flatten (repeat b a)))
          (and (string? a) (integer? b)) (reduce str (repeat b a))))

  (defn findAll [coll clos]
    (reduce conj [] (filter clos coll)))
Hope reducers give a speed boost by removing the laziness I'm not using.



Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact

Search: