5 ms·

Clojure is slower than Java.

• simias 13y ago
  But is it slower than the equivalent functional code ported to Java?
  • glibgil 13y ago
    Yes, of course it is. Nothing in Clojure is faster than Java, not even functional programming.
    • chongli 13y ago
      Nothing in Clojure is faster than Java, not even functional programming. Right, but that's completely ignoring programmer efficiency (or sanity). Trying to do the things that Clojure was designed for in Java is a recipe for pain.
      • glibgil 13y ago
        No, it works fine. If you are doing FP then things stay well organized no matter the language.
        • chongli 13y ago
          No, I'm talking about the things which Clojure is really good at. Macros, embedded domain-specific languages etc. Try doing something like core.logic in pure Java.
          • glibgil 13y ago
            core.logic can be done cleanly in Java, no problem. For example, here is an FP library in pure Java. Nice clean code. Easy to use. https://code.google.com/p/totallylazy/ https://code.google.com/p/totallylazy/
            • chongli 13y ago
              That is not a demonstration of what core.logic does.
              • glibgil 13y ago
                Reread my comment. And then again. Did I claim that? What did I claim? Now reread again.
    • Tuna-Fish 13y ago
      If you are using Clojure's built-in types to do fp vs Java's built-in types, Clojure will win by an absurd margin.
      • glibgil 13y ago
        You don't need Java's built-in types when you are doing FP in Java. In fact, it is not really FP if you do. Java FP is faster than Clojure in everything.
    • simias 13y ago
      Uh, I didn't expect that. How come? Does it mean that writing a Clojure-to-Java transpiler would yield better results than using the "native" Clojure implementation? It seems backwards to me that an implementation designed specifically for functional programming fails to beat the more "generic" (or arguably OOP-oriented) language on its own turf.
      • chii 13y ago
        > Does it mean that writing a Clojure-to-Java transpiler would yield better results than using the "native" Clojure implementation? yea, it does sound weird, because if a closure-to-java compiler yields better results, then why not just use that as the closure compiler? I agree that hand writing a piece of code in java that did the exact same thing as a piece of closure code would be faster in java than closure, because the programmer could make shortcuts where he knows he could.