# programming hard or hardly programming

Posted on April 4, 2013

I just found this article on The Daily WTF. After reading it, I began to think to myself: “Am I really choosing the right language / environment for the apps i might write?” and “Am I simply doing all this right?”

On the one hand, Java runs on (damn near) anything, so running it on a Windows machine (should be) just as easy as getting it to run on a Linux machine. On the other hand, there are databases to consider, and from all that I’ve heard, running PostgreSQL or MySQL on a Windows machine - while not impossible - is difficult. So now I’m torn, keep learning Clojure, or learn something else. Also, I would need to find a decent database, thats cheap (aka free) that runs on Windows & has no limitations (or at least the same limitations as say, MySQL). Hostly I don’t know any database very well, so PostgreSQL, MySQL (or MariaDB), MongoDB, CouchDB, or even learning Access DB would be 9/10 the same. The functionality is different, sure, but I need to keep things small in order to be sure that I can handle everything. And even though I might be able to think about & administer the system, I may need to hand it off to somebody at some later time, so keeping the code as readable & open as possible is best. I mean, I love the guys at my shop, so I don’t want to leave them in a situation where they have to hire an “expert Objective D# programmer” just to make some changes to their app. F# may be just as cool (if not cooler than) Clojure, but would they be better served if i wrote it all in that or C#? Or something else entirely???

There is also the question of “Am I doing this right?” This is something I often think about when I’m programming: “Am I doing this in the most effective way that gets across my point and wont confuse me (or someone else) later?” Sometimes I do it wrong, like in this instance (taken from my Clojure and Rosalind project):

 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24  (ns rosalind.basics) (defn dna-counter-fn "This actually does the counting" [dna cnt] (loop [d dna c cnt] (let [l (first d) r (rest d)] (case l \A (def h (assoc c :A (+ 1 (:A c)))) \G (def h (assoc c :G (+ 1 (:G c)))) \C (def h (assoc c :C (+ 1 (:C c)))) \T (def h (assoc c :T (+ 1 (:T c)))) nil) (if (empty? r) h (recur r h))))) (defn dna-count "Counts DNA Nucleotides" [dna] (let [h (dna-counter-fn dna {:A 0 :C 0 :T 0 :G 0})] [(:A h) (:C h) (:G h) (:T h)]))

My code went from that, to this (when I remembered that fold & reduce are staples in functional languages…):

 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15  (ns rosalind.basics) (defn count-dna "Counts DNA Nucleotides" [dna-str] (reduce (fn [c base] (case base \A (assoc c :A (+ 1 (:A c))) \G (assoc c :G (+ 1 (:G c))) \C (assoc c :C (+ 1 (:C c))) \T (assoc c :T (+ 1 (:T c))) nil)) {:A 0 :C 0 :T 0 :G 0} dna-str)) ;; ...

The second is definitely shorter, and for anyone who knows how to program in that language, they have a better sense of what the code is doing in the latter, as compared to the former. I don’t mind being wrong. In fact, so much so that pretty much all of the code that I’ve ever written is up on GitHub and I’m not afraid of anyone going through the history. In fact, if you want to see those 2 commits you can go here for the first and a link for the refactoring. (again, you should only really browse the code if you’re not going to solve the Project Rosalind stuff or you’ve already solved it, because there are answers there… exposed!) There is a little more to it than is in that commit, mainly the function no longer returns a vector but a map, but in the end it is easier to read & more comprehensible to other developers. So, if i were to build this big spiffy app in Clojure or C# or JavaScript or INTERCAL would it be as comprehensible as possible to my future self or someone there to replace me?

…maybe I should get a “replies” section to my blog so people can post stuff in response to my ramblings… or you can just head over to my Google+ page & respond there: Justin P.