more notes on using cljr

For work at CompassLabs, I now use the Clojure setup that everyone else uses (emacs+swank-clojure, with our custom repositories). For my own Clojure hacking (my own projects) I have just about settled on using cljr for convenience and agility. For me, the big win is being able to access Clojure libraries, Java libraries, and JAR files containing data sets I use often for NLP work from any directory. I don't need a heavy weight project, like for example, using Leiningen with all dependencies locally loaded. cljr uses Leiningen to manage the packages in the single ~/.cljr repository. When you startup cljr, everything in ~/.cljr is on your JVM classpath: this may seem a little heavy, but it is very convenient.

As an example, this morning I noticed an old Twitter direct message from the author of Nozzle library asking me if I had a chance to try it. Instead of setting up a separate Leiningen project directory, I just did a cljr install com.ashafa/nozzle 0.2.1, went to my catch-all directory where I keep short snippets of Clojure code, and entered Tunde's test program for Nozzle:

;; assumes: cljr install com.ashafa/nozzle 0.2.1

(use 'com.ashafa.nozzle)

(def username (System/getenv "TWITTER_ACCOUNT"))
(def passwd   (System/getenv "TWITTER_PASSWD"))

(defn my-callback 
   [message] 
   (println message))

(def noz (create-nozzle "filter" username passwd my-callback {:track "twitter"}))

and running it is as simple as:

cljr run nozzle-twitter-test.clj

or, if you want to use emacs+swank:

cljr swank

and after starting Emacs in the same directory:

M-x slime-connect

and in the repl:

(load "nozzle-twitter-test")

I still use IntelliJ with the Clojure plugin for some projects also.