Introduction to Clojure given as a tech talk at my work

1. Clojure
The LISP that makes the JVM dynamic
http://clojure.org

2. What is Clojure?
Clojure is a dynamic, LISP-like programming
language that runs on the JVM.

3. Clojure History
Started in 2007
Original author (and BDFL) Rich Hickey
Currently version 1.1 (1.2 expected within the next
month)
Available under Eclipse Public License
“The Eclipse Public License is designed to be a
business-friendly free software license and
features weaker copyleft provisions than
contemporary licenses such as the GNU General
Public License (GPL).”

4. Defining features
LISP syntax
Macros
Immutability
Functional programming
Concurrency
Java interoperability
REPL-based

5. Syntax
numbers - 1234
ratios - 22/7
strings - “foo”, “bar”
characters - a b c
symbols - foo, bar
keywords - :foo, :bar
boolean - true/false
null - nil

6. Syntax
Lists - (1 2 3 4)
Vectors - [1 2 3 4]
Maps - {:a 1 :b 2 :c 3} or {:a 1, :b 2, :c 3}
Sets - #{foo bar baz}

7. Syntax
That’s it. There is no other syntax[1].
Data structures are code.
Homoiconic
All data literals stand for themselves
except:
Lists
Symbols
[1]: Technically, there are other special symbols for shortcutting syntax provided for the developer, but no other necessary syntax

8. Lists
(+ 1 1) ; => 2
(+ 1 2 3) ; => 6
(+ 1 2 (+ 3 4)) ; => 10
(class “foo”) ; => java.lang.String
(first [:a :b :c]) ; => :a
(rest ‘(“foo” “bar” “baz”)) ; => (“bar” “baz”)
The ‘ tells the reader not to evaluate the first element as a function
‘(1 2 3) is shorthand for (quote (1 2 3))

9. Hello World
(ns helloworld)
(defn hello
“An example function”
[string]
(println “Hello” string))
(hello “World”) ; => “Hello World”

10. REPL demo

11. Operating on
Collections
Collections are the main datastructure in
Clojure
A collection can be a list, a vector, a map or a
set
Some example collection functions:
first, second, last, count, reverse, concat,
conj, contains?, map, reduce, apply, filter,
some, remove, every?, nth, into, doall,
repeat, repeatedly, range,

12. Laziness
Sequences can be lazy, which allows them to be
infinite:
(cycle [1 2 3]) ; => (1 2 3 1 2 3 1 2 3 ...)
(repeat :a) ; => (:a :a :a :a :a :a :a ...)
When using lazy sequences, processing does not
occur until the sequence is realized.
(class (repeat :a)) ; => clojure.lang.LazySeq
Realize (part of) a lazy sequence:
(take 5 (repeat 42)) ; => (42 42 42 42 42)

13. More laziness
Can do neat things with infinite
sequences:
(filter even? (iterate inc 1)) ; => (2 4 6 8...)
Lazily read lines from a file:
(read-lines “/tmp/foo.txt”)
Lines aren’t read until they are used in
the returned list

14. Java interop
(.toLowerCase “EMC”) ; => “emc”
(<method> <object> [<arg1> <arg2> ...])
(StringBuffer. “foo” ) ; => #<StringBuffer foo>
(Integer/parseInt “5”) ; => 5
(.contains “My spoon is too big.” “spoon“) ; => true
(javax.swing.JOptionPane/showMessageDialog nil "Sup!")
“Clojure is a better Java than Java”

15. Java datastructures
Anything that’s a Collection (or array) in Java can easily be
treated (or converted) as a Clojure collection
(first (.split “This is a sentence” “ “)) ; => “This”
(import ‘java.util.ArrayList)
(def foo (ArrayList. [1 2 3])) ; => #<ArrayList [1, 2, 3]>
(seq foo) ; => (1 2 3)
(vec foo) ; => [1 2 3]

16. More Java-interop
Type hints
(defn #^String uppercase [#^String s] (.toUpperCase s))
Proxying Java classes
(doto (javax.swing.JFrame.)
(addKeyListener (proxy [java.awt.event.KeyListener] []
(keyPressed [e] (println (.getKeyChar e) " key pressed"))
(keyReleased [e] (println (.getKeyChar e) " key released"))
(keyTyped [e] (println (.getKeyChar e) " key typed"))))
(setVisible true))
Lots more
gen-class, annotations, primitives, definterface, defprotocol

17. Immutability
(defn add-age
[person]
(assoc person :age 10))
(def timmy {:name “Timmy”})
(print timmy) ; => {:name “Timmy”}
(add-age timmy) ; => {:name “Timmy”, :age 10}
; timmy never changes
(print timmy) ; => {:name “Timmy”}
Under the hood: 32-way trie trees. O(log32n)

18. Functional
Programming
Functions are intended to be side-effect free
They take values and return values
Same argument, same result
No notion of time
Easy to add concurrency to the mix sanely
Testing is easier because there is no state
Having no side effects isn’t always possible (io,
mutable java objects)

19. Concurrency
Locks are bad
Clojure’s defining concurrency feature is STM
Gives you ACI out of ACID
Atomicity
Consistency
Isolation
Immutability is what makes this possible
4 methods for changing data concurrently:

20. Dead-simple
Concurrency
(def accounts [{:name “Bob” :balance 1000}
{:name “Sue” :balance 2000}
{:name “Joe” :balance 3000}
{:name “Ann” :balance 4000}])
(defn add-interest
[acct]
(let [bal (:balance acct)]
(Thread/sleep 1500)
(assoc acct :balance (* bal 1.25))))
(time (doall (map add-interest accounts)))
; => "Elapsed time: 6001.023 msecs"
(time (doall (pmap add-interest accounts)))
; => "Elapsed time: 1519.73 msecs"

21. More simple
concurrency
All functions defined in Clojure implement
Runnable:
(defn dowork [] (println “doing work”))
(.start (Thread. dowork)) ; => “doing work”
Access to all of Java’s concurrency stuff
(ThreadPoolExecutor, etc) with Java
interop

22. Concurrency (Refs)
(def x (ref 100))
@x ; => 100
(def amounts [5 10 1 9 2])
(defn dec-x
[n]
(dosync
(alter x #(- % n))))
(pmap dec-x amounts)
@x ; => 73

23. Concurrency (Atoms)
(def x (atom 100))
@x ; => 100
(def amounts [5 10 1 9 2])
(defn dec-x
[n]
(swap! x #(- % n)))
(pmap dec-x amounts)
@x ; => 73

24. Concurrency (Agents)
(def x (agent 0))
@x ; => 0
(defn increment [c n] (+ c n))
(send x increment 5) ; @x -> 5
(send x increment 10) ; @x -> 15

25. Macros
(defmacro def-name-filter
[n r]
(let [docstring (str "Given a list of people maps, filter " n " in a list.")]
`(defn ~n
~docstring
[elements#]
(filter (fn [t#] (re-find ~r (:name t#))) elements#))))
(def-name-filter foo-filter #"^foo")
(def people [{:name "pat"} {:name "foobar"} {:name "foo"} {:name "bar"}])
(doc foo-filter)
; => Given a list of people maps, filter foo-filter in a list.
(foo-filter people)
; => ({:name "foobar"} {:name "foo"})

26. Macroexpand example

27. List de-structuring
; Destructure in defn
(defn print-items
[item & rest]
; Destructure in let (println "item:" item)
(println "rest:" rest)
(def my-list ‘([:a 1] [:b 2] [:c 3])) (if-not (nil? rest)
(apply print-items rest)))
(defn print-pair
[pair] (print-items 1 2 3 4 5)
(let [[key val] pair]
(println "key:" key "val:" val))) item: 1
rest: (2 3 4 5)
(map print-pair my-list) item: 2
rest: (3 4 5)
key: :a val: 1 item: 3
key: :b val: 2 rest: (4 5)
key: :c val: 3 item: 4
rest: (5)
item: 5
rest: nil

28. Much More
Polymorphism/multimethods
Pre and Post conditions for functions
Futures, Promises
Watchers
Bindings/Transients
Metadata features
Compilation
Clojure-in-Clojure
So much more!

29. More info
http://clojure.org
http://clojure.blip.tv
http://www.assembla.com/wiki/show/clojure/Getting_Started
#clojure on irc.freenode.net
Come talk to me about it!

30. Thanks!
Questions?