Clojure: Functional Concurrency for the JVM (presented at Open Source Bridge)

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    7 Favorites & 1 Event

    Clojure: Functional Concurrency for the JVM (presented at Open Source Bridge) - Presentation Transcript

    1. Clojure: Functional Concurrency for the JVM Howard M. Lewis Ship Director of Open Source Technology Formos Software Development howard.lewis.ship@formos.com © 2009 Formos Software Development
    2. Clojure: The Language © 2009 Formos Software Development
    3. http://xkcd.com/297/ Clojure: The Language © 2009 Formos Software Development
    4. Rich Hickey © 2009 Formos Software Development
    5. Code is Data Quoted list of '(1 2 3) numbers (biggest 5 42) Function call Function definition (defn biggest "Find the maximum of two numbers" [x y] (if (> x y) x y)) © 2009 Formos Software Development
    6. Read Eval Print Loop user=> (defn biggest "Find the maximum of two numbers" [x y] (if (> x y) x y)) #=(var user/biggest) user=> (biggest 5 42) 42 user=> (doc biggest) ------------------------- user/biggest ([x y]) Find the maximum of two numbers nil user=> '(1 2 3) (1 2 3) user=> '(biggest 5 42) (biggest 5 42) user=> (first '(biggest 5 42)) biggest user=> (eval '(biggest 5 42)) 42 © 2009 Formos Software Development
    7. There Is No Interpreter Source Code Repl Input Clojure User Classes Java Evaluator Compiler Clojure Source Files Java Libraries JVM Operating System © 2009 Formos Software Development
    8. Clojure Literals user=> 42 42 user=> "A Clojure String" "A Clojure String" user=> nil nil user=> :balance :balance user=> true true user=> false false © 2009 Formos Software Development
    9. Clojure Literals user=> 5 5 user=> 5.001 5.001 user=> 22/7 22/7 user=> (* 2 22/7) 44/7 user=> (* 100000 100000 100000) 1000000000000000 user=> (+ 5. 0.000000000000000001) 5.0 user=> (+ 5.0M 0.000000000000000001M) 5.000000000000000001M © 2009 Formos Software Development
    10. Java Interop factory.setNamespaceAware(true) (.setNamespaceAware factory true) new StringBuffer() (new StringBuffer) (StringBuffer.) factory.newSAXParser().parse(src, handler) (.. factory newSAXParser (parse src handler)) MyObject.ivar = "foo"; (set! (. MyObject ivar) "foo") © 2009 Formos Software Development
    11. Java Interop frame.add(panel, BorderLayout.CENTER); frame.add(greetButton, BorderLayout.SOUTH); frame.pack(); frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE); frame.setVisible(true); (doto frame (.add panel BorderLayout/CENTER) (.add greet-button BorderLayout/SOUTH) (.pack) (.setDefaultCloseOperation JFrame/EXIT_ON_CLOSE) (.setVisible true)) © 2009 Formos Software Development
    12. Clojure Collections: Lists lst user=> (def lst `(1 2 3)) #=(var user/lst) user=> lst (1 2 3) user=> (first lst) 1 1 user=> (rest lst) (2 3) user=> (conj lst 4) 2 (4 1 2 3) user=> (cons 4 lst) (4 1 2 3) 3 © 2009 Formos Software Development
    13. Clojure Collections: Lists 4 lst user=> (def lst `(1 2 3)) #=(var user/lst) user=> lst (1 2 3) user=> (first lst) 1 1 user=> (rest lst) (2 3) user=> (conj lst 4) 2 (4 1 2 3) user=> (cons 4 lst) (4 1 2 3) 3 © 2009 Formos Software Development
    14. Clojure Collections: Vectors user=> (def v [:moe :larry :curly]) #=(var user/v) user=> v [:moe :larry :curly] user=> (first v) :moe user=> (rest v) (:larry :curly) user=> (conj v :shemp) [:moe :larry :curly :shemp] user=> (cons :shemp v) (:shemp :moe :larry :curly) user=> v [:moe :larry :curly] user=> (v 1) :larry vector is a function of its indexes © 2009 Formos Software Development
    15. Clojure Collections: Map user=> (def m {:first-name "Howard" :last-name "Lewis Ship"}) #=(var user/m) user=> m {:last-name "Lewis Ship", :first-name "Howard"} user=> (get m :last-name) "Lewis Ship" map is a user=> (m :last-name) function of "Lewis Ship" its keys user=> (assoc m :company "Formos") {:company "Formos", :last-name "Lewis Ship", :first-name "Howard"} user=> m {:last-name "Lewis Ship", :first-name "Howard"} user=> (:first-name m) "Howard" user=> (:ssn m) nil Keywords are functions, too! © 2009 Formos Software Development
    16. Clojure Collections: Sets user=> (def s #{"Howard" "Suzanne" "Molly" "Jim"}) #=(var user/s) user=> s #{"Howard" "Jim" "Molly" "Suzanne"} user=> (contains? s "Howard") true user=> (contains? s "howard") false set is a user=> (s "Howard") function of "Howard" its elements user=> (s "Rhys") nil user=> (conj s "Howard") #{"Howard" "Jim" "Molly" "Suzanne"} user=> (conj s "Scott") #{"Howard" "Jim" "Molly" "Suzanne" "Scott"} © 2009 Formos Software Development
    17. © 2009 Formos Software Development
    18. ❝For alumni of other languages, beginning to use Lisp may be like stepping onto a skating rink for the first time. It’s actually much easier to get around on ice than it is on dry land—if you use skates. Till then you will be left wondering what people see in this sport.❞ Paul Graham © 2009 Formos Software Development
    19. Functional Programming © 2009 Formos Software Development
    20. Functional Programming © 2009 Formos Software Development
    21. © 2009 Formos Software Development
    22. No Mutable State© 2009 Formos Software Development
    23. © 2009 Formos Software Development
    24. No Side Effects © 2009 Formos Software Development
    25. © 2009 Formos Software Development
    26. First Class Functions © 2009 Formos Software Development
    27. © 2009 Formos Software Development
    28. Functional Composition © 2009 Formos Software Development
    29. Functional Programming in Java public void saveOrUpdate(final Employee employee) { HibernateCallback callback = new HibernateCallback() { public Object doInHibernate(Session session) throws HibernateException,SQLException { session.saveOrUpdate(employee); return null; } }; hibernateTemplate.execute(callback); } Outer function controls the SwingUtilities.invokeLater(new Runnable() context: { public void run() • Thread { • Exception handling progressBar.setValue(progressBar.getValue() + 1); } • Parameters }); © 2009 Formos Software Development
    30. Functional Java Collections public interface Predicate<T> { boolean accept(T value); } public static <T> Collection<T> filter(Predicate<T> pred, Collection<T> coll) { Collection<T> out = new ArrayList<T>(); for (T item : coll) { if (pred.accept(item)) out.add(item); } return out; } return CollectionUtils.filter(new Predicate<String>() { public boolean accept(String value) { return !value.startsWith("."); } }, names); © 2009 Formos Software Development
    31. Functional Clojure Collections Function Anonymous parameter function (filter #(not (.startsWith % ".")) names) Member access form user=> (def names ["fred" "barney" ".hidden" "wilma"]) #=(var user/names) user=> (filter #(not (.startsWith % ".")) names) ("fred" "barney" "wilma") user=> (remove #(.startsWith % ".") names) ("fred" "barney" "wilma") user=> © 2009 Formos Software Development
    32. First Class Functions (filter #(not (.startsWith % ".")) names) function as parameter to function (defn require-extension [ext] (fn [file-name] (= ext (last (split-string file-name "."))))) function as return value (defn filter-by-extension [ext coll] (filter (require-extension ext) coll)) composing functions © 2009 Formos Software Development
    33. Bridging Java and Clojure SwingUtilities.invokeLater(new Runnable() { public void run() { progressBar.setValue(progressBar.getValue() + 1); } }); Invoke static method (SwingUtilities/invokeLater #(.setValue progressBar (inc (.getValue progressBar)))) Clojure functions implement: • Runnable • Callable • Comparator © 2009 Formos Software Development
    34. Life without the for loop public static int sum(int[] vals) { int total = 0; col for (int val : vals) total += val; return total; } x y (defn sum z [col] (reduce + 0 col)) ➠ 0 + col[0] + col[1] + col[2] ... © 2009 Formos Software Development
    35. Life without the for loop public static int sum(int[] vals) { int total = 0; col for (int val : vals) total += val; return total; } x y (defn sum z [col] (reduce + 0 col)) ➠ 0 + col[0] + col[1] + col[2] ... ➠ (+ 0 (first col)) (+ (+ 0 (first col)) (first (rest col))) (+ (+ (+ 0 (first col)) (first (rest col))) (first (rest (rest col)))) ... © 2009 Formos Software Development
    36. Life without the for loop public static String[] formatDoubles(double[] inputs) { String[] output = new String[inputs.length]; for (int i = 0; i < input.length; i++) output[i] = String.format("%9.2f", inputs[i]); return output; } (defn format-doubles f [col] (map #(format "%9.2f" %) col)) f Apply function to each item, forming f new seq user=> (format-doubles '(2.5 3.7 -22.7)) (" 2.50" " 3.70" " -22.70") © 2009 Formos Software Development
    37. for: list comprehension user=> (range 0 5) (0 1 2 3 4) user=> (for [x (range 0 10) :when (even? x)] x) (0 2 4 6 8) user=> (for [x (range 1 5) y (range 0 x)] [x y]) ([1 0] [2 0] [2 1] [3 0] [3 1] [3 2] [4 0] [4 1] [4 2] [4 3]) © 2009 Formos Software Development
    38. for: list comprehension user=> (range 0 5) (0 1 2 3 4) user=> (for [x (range 0 10) :when (even? x)] x) (0 2 4 6 8) user=> (for [x (range 1 5) y (range 0 x)] [x y]) ([1 0] [2 0] [2 1] [3 0] [3 1] [3 2] [4 0] [4 1] [4 2] [4 3]) (defn convert-attributes-to-tokens [attrs] (for [x (range (.getLength attrs))] (let [uri (.getURI attrs x) name (.getLocalName attrs x) value (.getValue attrs x) token (…) ] token))) © 2009 Formos Software Development
    39. Laziness is a Virtue Image © 2007 Jon Fife http://flickr.com/photos/good-karma/577632972/ © 2009 Formos Software Development
    40. Laziness is a Virtue user=> (take 20 (iterate inc 1)) (1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20) user=> (take 20 (map * (iterate inc 1) (iterate inc 1))) (1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361 400) Image © 2007 Jon Fife http://flickr.com/photos/good-karma/577632972/ © 2009 Formos Software Development
    41. Laziness © 2009 Formos Software Development
    42. Java: Data Encapsulated in Objects Person Person Person Person firstName: "Howard" firstName: "Scott" firstName: "Molly" firstName: "David" lastName: "Lewis Ship" lastName: "Simon" lastName: "Newman" lastName: "Goldman" age: 42 age: 44 age: 29 age: 42 public double averageAge(Collection<Person> persons) { double total = 0.0; for (Person p : persons) total += p.getAge(); return total / persons.size(); } © 2009 Formos Software Development
    43. Clojure: Data in Transformable Collections :first-name Howard :first-name Scott :first-name Molly { :last-name Lewis Ship } { :last-name Simon }{ :last-name Newman } :age 42 :age 44 :age 29 user=> persons [{:first-name "Howard", :last-name "Lewis Ship", :age 42} {:first-name "Scott", :last-name "Simon", :age 44} {:first-name "Molly", :last-name "Newman", :age 29}] user=> (map :age persons) (42 44 29) user=> (apply + (map :age persons)) 115 user=> © 2009 Formos Software Development
    44. Clojure: Data in Transformable Collections :first-name Howard :first-name Scott :first-name Molly { :last-name Lewis Ship } { :last-name Simon }{ :last-name Newman } :age 42 :age 44 :age 29 user=> persons [{:first-name "Howard", :last-name "Lewis Ship", :age 42} {:first-name "Scott", :last-name "Simon", :age 44} {:first-name "Molly", :last-name "Newman", :age 29}] user=> (map :age persons) (42 44 29) user=> (apply + (map :age persons)) 115 user=> (defn avg-age [coll] (/ (apply + (map :age coll)) (count coll))) © 2009 Formos Software Development
    45. Clojure: Data in Transformable Collections :first-name Howard :first-name Scott :first-name Molly { :last-name Lewis Ship } { :last-name Simon }{ :last-name Newman } :age 42 :age 44 :age 29 user=> persons [{:first-name "Howard", :last-name "Lewis Ship", :age 42} {:first-name "Scott", :last-name "Simon", :age 44} {:first-name "Molly", :last-name "Newman", :age 29}] user=> (map :age persons) (42 44 29) user=> (apply + (map :age persons)) 115 user=> (defn avg-age [coll] (/ (apply + (map :age coll)) (count coll))) (defn avg (defn avg-age [f coll] [coll] (/ (apply + (map f coll))) (count coll))) (avg :age coll)) © 2009 Formos Software Development
    46. Clojure: Data in Transformable Collections :first-name Howard :first-name Scott :first-name Molly { :last-name Lewis Ship } { :last-name Simon }{ :last-name Newman } :age 42 :age 44 :age 29 user=> persons [{:first-name "Howard", :last-name "Lewis Ship", :age 42} {:first-name "Scott", :last-name "Simon", :age 44} {:first-name "Molly", :last-name "Newman", :age 29}] user=> (map :age persons) (42 44 29) user=> (apply + (map :age persons)) 115 user=> (defn avg-age [coll] (/ (apply + (map :age coll)) (count coll))) (defn avg (defn avg-age [f coll] [coll] (/ (apply + (map f coll))) (count coll))) (avg :age coll)) (avg #(count (:last-name %)) persons) © 2009 Formos Software Development
    47. © 2009 Formos Software Development
    48. ❝Somehow the idea of reusability got attached to object-oriented programming in the 1980s, and no amount of evidence to the contrary seems to be able to shake it free.❞ Paul Graham © 2009 Formos Software Development
    49. Clojure Concurrency © 2009 Formos Software Development
    50. Solving Deadlocks: Timeout & Retry © 2009 Formos Software Development
    51. Solving Deadlocks: Specific Lock Order © 2009 Formos Software Development
    52. Solving Deadlocks: Coarse Locks Image © 2008 Marcin Wichary http://flickr.com/photos/mwichary/2222776430/ © 2009 Formos Software Development
    53. Locks are the Enemy Image © 2007 James Manners http://flickr.com/photos/jmanners/443421045/ © 2009 Formos Software Development
    54. Clojure: Software Transactional Memory (def savings (ref 1000.)) (def checking (ref 2000.)) (def mm (ref 7000.)) (defn transfer "Transaction to transfer money from one account to another." [from to amount] (dosync (alter from - amount) (alter to + amount))) (transfer checking savings 500.) (transfer mm checking 300.) - 500. + 500. - 300. + 300. Checking 1000. 500. 500. Savings 2000. 2500. 2800. 2800. MM 7000. 6700. 6700. © 2009 Formos Software Development
    55. Retries: Transactions Are Speculative (def savings (ref 1000.)) (def checking (ref 2000.)) (def mm (ref 7000.)) (defn transfer "Transaction to transfer money from one account to another." [from to amount] (dosync (alter from - amount) (alter to + amount))) (transfer checking savings 500.) (transfer mm checking 300.) + 300. - 500. - 300. + 500. - 300. + 300. Checking 1000. 500. X 500. Savings 2000. 2500. 2300. 2500. 2800. 2800. MM 7000. 6700. 7000. 6700. 6700. © 2009 Formos Software Development
    56. © 2009 Formos Software Development
    57. No Blocking © 2009 Formos Software Development
    58. © 2009 Formos Software Development
    59. No Locks © 2009 Formos Software Development
    60. © 2009 Formos Software Development
    61. Persistent Collections © 2009 Formos Software Development
    62. Managing Mutation • What can change? • Reference types: atom, var, agent, ref • When can they change? • When are changes visible to other threads? Image © 2008 Daniel Chan http://flickr.com/photos/chanchan222/2847443980/ © 2009 Formos Software Development
    63. Atoms • Shared, Global • Changes are atomic, synchronous, & non-blocking • (swap!): Pass value to function yielding new value • (reset!): Force new value, regardless of existing value user=> (def queue (atom [])) #'user/queue user=> @queue [] user=> (swap! queue conj {:parse "http://www.clojure.org/"}) [{:parse "http://www.clojure.org/"}] user=> @queue [{:parse "http://www.clojure.org/"}] user=> (reset! queue []) [] user=> @queue [] user=> © 2009 Formos Software Development
    64. Vars — Per-Thread Mutables • (def) sets global binding • (binding) to set up a per-thread override Image © 2005 Jack Keene http://www.flickr.com/photos/whatknot/3118124/ • (set!) if per-thread binding user=> (def x 1) #=(var user/x) user=> x 1 user=> (defn manipulate-x [] (binding [x 2] (printf "Local x is %d" x) (set! x 3) (printf "nLocal x is now %dn" x))) #=(var user/manipulate-x) user=> (.run (Thread. manipulate-x)) Local x is 2 Local x is now 3 nil user=> x 1 © 2009 Formos Software Development
    65. Interfacing with Java APIs (def *tokens*) (defn add-token [token] (set! *tokens* (conj *tokens* token))) (def sax-handler (proxy [DefaultHandler] [] (startElement [uri local-name q-name attrs] (flush-text) (add-token …)) … )) (defn tokenize-xml [src] (binding [*tokens* []] (let [factory (SAXParserFactory/newInstance)] (.setNamespaceAware factory true) (.. factory newSAXParser (parse src sax-handler)) *tokens*))) © 2009 Formos Software Development
    66. Everything's a Var! Function Namespace Var Value ... Symbol such as x or map © 2009 Formos Software Development
    67. Functions are stored in Vars user=> (defn say-hello [] (println "Hello")) #'user/say-hello user=> (say-hello) Hello nil user=> (binding [say-hello #(println "Goodbye")] (say-hello)) Goodbye nil user=> (say-hello) Hello nil user=> © 2009 Formos Software Development
    68. Agents — Single Thread Writes user=> (def savings (agent 1000.)) #=(var user/savings) user=> (def checking (agent 2000.)) #=(var user/checking) user=> @savings 1000 user=> @checking 2000 user=> (send savings - 300.) #<clojure.lang.Agent@c3233b> user=> (send checking + 300.) #<clojure.lang.Agent@67e5a7> user=> @savings 700 user=> @checking 2300 •Asynchonous •Single threaded •Non-transactional © 2009 Formos Software Development
    69. Refs — Software Transactional Memory (def savings (ref 1000.)) (def checking (ref 2000.)) (def mm (ref 7000.)) (defn transfer "Transaction to transfer money from one account to another." [from to amount] (dosync (alter from - amount) (alter to + amount))) (transfer checking savings 500.) (transfer mm checking 300.) user=> @checking 2000 user=> @savings 1000 user=> (transfer checking savings 500.) 1500 user=> @checking 1500 user=> @savings 1500 user=> (ref-set savings 2000.) java.lang.IllegalStateException: No transaction running (NO_SOURCE_FILE:0) © 2009 Formos Software Development
    70. Concurrency Notes • All reference types can have a validator function • All refs can have a watcher: an agent notified of changes • (send) inside (dosync) waits until successful completion • No guarantees when calling Java objects © 2009 Formos Software Development
    71. © 2009 Formos Software Development
    72. ❝The key to performance is elegance, not battalions of special cases.❞ Jon Bentley and Doug McIlroy © 2009 Formos Software Development
    73. Wrap Up © 2009 Formos Software Development
    74. Clojure • 1.0 release: May 4 2009 • Simple, regular syntax • Improves on Lisp: vectors, maps, sets • Fully integrates with Java http://www.clojure.org • Impressive functional & concurrency support • Many features not covered here © 2009 Formos Software Development
    75. Stuart Halloway Pragmatic Bookshelf http://pragprog.com/titles/shcloj/programming-clojure © 2009 Formos Software Development
    76. http://jnb.ociweb.com/jnb/jnbMar2009.html © 2009 Formos Software Development
    77. Object Oriented Copyright © A. Lipson 2003 Copyright © 2007 Alan Chia http://www.andrewlipson.com/escher/relativity.html http://flickr.com/photos/seven13avenue/2080281038/ © 2009 Formos Software Development
    78. Object Oriented Copyright © A. Lipson 2003 Copyright © 2007 Alan Chia http://www.andrewlipson.com/escher/relativity.html http://flickr.com/photos/seven13avenue/2080281038/ © 2009 Formos Software Development
    79. Functional © 2009 Formos Software Development
    80. Functional Image © 2007 Woodley Wonderworks http://flickr.com/photos/wwworks/2222523486/ © 2009 Formos Software Development

    + Howard Lewis ShipHoward Lewis Ship, 5 months ago

    custom

    1663 views, 7 favs, 2 embeds more stats

    Slides from my 45 minute talk on Clojure presented more

    More info about this document

    © All Rights Reserved

    Go to text version

    • Total Views 1663
      • 1426 on SlideShare
      • 237 from embeds
    • Comments 0
    • Favorites 7
    • Downloads 34
    Most viewed embeds
    • 236 views on http://tapestryjava.blogspot.com
    • 1 views on http://feeds2.feedburner.com

    more

    All embeds
    • 236 views on http://tapestryjava.blogspot.com
    • 1 views on http://feeds2.feedburner.com

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories

    Groups / Events