Deck used for the Fun with Automata talk at Frontier Developers, May 24 2012. Accompanying material at: https://github.com/jdegoes/fun-with-automata

1. Frontier Developers
6:00: Pizza & Networking
6:20: Presentation
https://github.com/jdegoes/fun-with-automata
WIFI: Scriblove / scribble
Thursday, May 24, 12

2. Fun with Automata!
https://github.com/jdegoes/fun-with-automata
John A. De Goes
@jdegoes http://precog.com
Thursday, May 24, 12

3. Automata
Abstract machines that transition between states
based on reading of input.
Transducers
Automata that read and write.
Deterministic Transducers
Output deterministically depends on current
state and prior input.
Thursday, May 24, 12

4. DTs vs Functions?
Function
A B
DT
A (B, DT)
Thursday, May 24, 12

5. “Functions with a
memory.”
Thursday, May 24, 12

6. Benefits
• Purely-functional yet stateful
• Incredibly composable
• Safely concurrent
• Recoverable
Thursday, May 24, 12

7. trait DT[A, B] {
def ! (input: A): (B, DT[A, B])
}
object DT {
def apply[A, B](f: A => (B, DT[A, B])) =
new DT[A, B] { def ! (v: A) = f(v) }
}
Thursday, May 24, 12

8. val d1 = {
def f(b: Boolean): DT[Int, Int] =
DT { v =>
val output = if (b) v * 2 else v
(output, f(!b))
}
f(false)
}
Thursday, May 24, 12

9. val (v1, d2) = d1 ! 1
// v1 = 1
val (v2, d3) = d2 ! 1
// v2 = 2
...
Thursday, May 24, 12

10. trait DT[A, B] {
def ! (input: A): (B, DT[A, B])
def !! (inputs: A*): (List[B], DT[A, B]) read many
def >>> [C] (that: DT[B, C]): DT[A, C] compose
def map[C](f: B => C): DT[A, C] output transform
def contramap[C](f: C => A): DT[C, B] input transform
def & [C, D](that: DT[C, D]): parallelize
DT[(A, C), (B, D)]
}
Thursday, May 24, 12

11. Helpers
def identity[A]: DT[A, A]
def constant[A, B](b: B): DT[A, B]
def rep[A]: DT[A, (A, A)]
def rep3[A]: DT[A, ((A, A), A)]
def merge[A, B, C](f: (A, B) => C): DT[(A, B), C]
def merge3[A, B, C, D](f: (A, B, C) => D): DT[((A, B), C), D]
Thursday, May 24, 12

12. Exercise 1
Develop DT and some of its helper methods in a
language of your choice.
Thursday, May 24, 12

13. Exercise 2
Develop a simple streaming analytics framework,
with aggregation and windowing.
Thursday, May 24, 12

14. Exercise 3
Create a recognizer for the language defined by
the grammar ab* (that is, one ‘a’ character
followed by zero or more ‘b’ characters). The
recognizer emits true’s until a character breaks
the pattern, then emits false’s continuously.
Thursday, May 24, 12