Brief history of “programming” Al-Khwarizmi, 860AD • “Programming” in the sense of devising rules , methods , procedures , recipes , algorithms , has always been present in human thought. • “Algorithm” originates from the name of Al-Khwarizmi, Arab mathematician from 800AD. “Algorists” 1504AD Sumerian division algorithm, 2000BC Sumerian division, translated Sumerian division, symbolically [The number is 4;10. What is its inverse? The number is x. What is its inverse? Proceed as follows. Proceed as follows. Form the inverse of 10, you will find 6. Form the inverse of y, you will find y’ . Multiply 6 by 4, you will find 24. Multiply y’ by z. You will find t. Add on 1, you will find 25. Add on 1. You will find u. Form the inverse of 25], you will find 2;24. Form the inverse of u. You will find u’ . [Multiply 2;24 by 6], you will find 14;24. Multiply u’ by y’ . You will find v. [The inverse is 14;24]. Such is the way to proceed. The inverse is v. Such is the way to proceed. 1 Euclid’s algorithm for GCD Functional vs imperative Proposition 2 Given two numbers not prime to one another, to find their greatest common • The Sumerian division algorithm, which is measure. representative of most algorithms in Mathematics, Let AB, CD be the two given numbers not prime to one another. is an example of a functional computation . Thus it is required to find the greatest common measure of AB, CD. • The Euclid’s GCD algorithm is an example of an ….. imperative computation, involving iteration and But, if CD does not measure AB, then the less of the numbers AB, CD, being state change. continually subtracted from the greater, some number will be left which will measure the one before it. • [For more details, see: Jean-Luc Chabert : A …Let such a number measure them, and let it be G. History of Algorithms , Springer, 1994] Now, since G measures CD, while CD measures BE, G also measures BE. Functional computation Imperative computation • Apply functions to input • Commands operate on the Store values or intermediate store , which contain values to compute new locations , and read or outputs. write them. • One view is: commands Commands • Type checking helps transform the state of the ensure the “plug- store ( i.e ., “functions” of compatibility” of the some sort). inputs/outputs. • Types are not of much help to ensure correctness. Language Evolution Lisp Algol 60 Algol 68 Introduction to Haskell Pascal C Smalltalk (Borrowed from Peyton Jones & John Mitchell) ML Modula C++ Haskell Java Many others: Algol 58, Algol W, Scheme, EL1, Mesa (PARC), Modula-2, Oberon, Modula-3, Fortran, Ada, Perl, Python, Ruby, C#, Javascript, F#… 2 C Programming Language ML programming language Dennis Ritchie, ACM Turing Award for Unix • Statically typed, general purpose systems programming • Statically typed, general-purpose programming language language – “Meta-Language” of the LCF theorem proving system • Computational model reflects underlying machine • Type safe, with formal semantics • Compiled language, but intended for interactive use • Relationship between arrays and pointers • Combination of Lisp and Algol-like features – An array is treated as a pointer to first element – Expression-oriented – E1[E2] is equivalent to ptr dereference: *((E1)+(E2)) – Higher-order functions – Pointer arithmetic is not common in other languages – Garbage collection • Not statically type safe – Abstract data types – Module system – If variable has type float, no guarantee value is floating pt – Exceptions • Ritchie quote • Used in printed textbook as example language – “C is quirky, flawed, and a tremendous success” Robin Milner, ACM Turing-Award for ML, LCF Theorem Prover, … Haskell Haskell B Curry • Haskell programming language is • Combinatory logic – Similar to ML: general-purpose, strongly typed, higher-order, – Influenced by Russell and Whitehead functional, supports type inference, interactive and compiled use – Developed combinators to represent – Different from ML: lazy evaluation, purely functional core, rapidly substitution – Alternate form of lambda calculus that evolving type system has been used in implementation structures • Designed by committee in 80’s and 90’s to unify research • Type inference efforts in lazy languages – Devised by Curry and Feys – Haskell 1.0 in 1990, Haskell ‘98, Haskell’ ongoing – Extended by Hindley, Milner – “A History of Haskell: Being Lazy with Class” HOPL 3 Although “Currying” and “Curried functions” Paul Hudak Simon are named after Curry, the idea was invented Peyton Jones by Schoenfinkel earlier John Hughes Phil Wadler Why Study Haskell? Why Study Haskell? • Good vehicle for studying language concepts • Functional programming will make you think • Types and type checking differently about programming. – General issues in static and dynamic typing – Mainstream languages are all about state – Type inference – Functional programming is all about values – Parametric polymorphism • Haskell is “cutting edge” – Ad hoc polymorphism (aka, overloading) • Control – A lot of current research is done using Haskell – Rise of multi-core, parallel programming likely to – Lazy vs. eager evaluation make minimizing state much more important – Tail recursion and continuations – Precise management of effects • New ideas can help make you a better programmer, in any language 3 Most Research Languages Successful Research Languages 1,000,000 1,000,000 10,000 10,000 Practitioners Practitioners 100 100 The slow death The quick death 1 1 Geeks Geeks 1yr 5yr 10yr 15yr 1yr 5yr 10yr 15yr C++, Java, Perl, Ruby Committee languages Threshold of immortality 1,000,000 1,000,000 10,000 10,000 Practitioners The complete Practitioners 100 absence of death 100 The slow death 1 1 Geeks Geeks 1yr 5yr 10yr 15yr 1yr 5yr 10yr 15yr “Learning Haskell is a great way of Haskell training yourself to think functionally so “I'm already looking at coding you are ready to take full advantage of problems and my mental C# 3.0 when it comes out” perspective is now shifting (blog Apr 2007) 1,000,000 back and forth between purely OO and more FP styled solutions” (blog Mar 2007) 10,000 Practitioners 100 The second life? 1 Geeks 1990 1995 2000 2005 2010 4 Function Types in Haskell Basic Overview of Haskell In Haskell, f :: A → B means for every x ∈ A, • Interactive Interpreter (ghci): read-eval-print ∈ f(x) = some element y = f(x) B run forever – ghci infers type before compiling or executing – Type system does not allow casts or other loopholes! • Examples Prelude> (5+3)-2 In words, “if f(x) terminates, then f(x) ∈ B.” 6 it :: Integer Prelude> if 5>3 then “Harry” else “Hermione” In ML, functions with type A → B can throw an “Harry” exception or have other effects, but not in Haskell it :: [Char] -- String is equivalent to [Char] Prelude> 5==4 False it :: Bool Overview by Type Simple Compound Types • Booleans True, False :: Bool Tuples if … then … else … --types must match (4, 5, “Griffendor”) :: (Integer, Integer, String) • Integers Lists 0, 1, 2, … :: Integer [] :: [a] -- polymorphic type +, * , … :: Integer -> Integer -> Integer 1 : [2, 3, 4] :: [Integer] -- infix cons notation • Strings “Ron Weasley” Records data Person = Person {firstName :: String, lastName :: String} • Floats hg = Person { firstName = “Hermione”, lastName = “Granger”} 1.0, 2, 3.14159, … --type classes to disambiguate Haskell Libraries Patterns and Declarations Functions and Pattern Matching • Patterns can be used in place of variables <pat> ::= <var> | <tuple> | <cons> | <record> … • Function declaration form <name> <pat > = <exp > • Value declarations 1 1 –General form: <pat> = <exp> <name> <pat 2> = <exp 2> … –Examples <name> <pat n> = <exp n> … myTuple = (“Flitwick”, “Snape”) • Examples (x,y) = myTuple myList = [1, 2, 3, 4] f (x,y) = x+y --argument must match pattern (x,y) length [] = 0 z:zs = myList length (x:s) = 1 + length(s) –Local declarations • let (x,y) = (2, “Snape”) in x * 4 5 More Functions on Lists More Efficient Reverse reverse xs = • Append lists let rev ( [], accum ) = accum rev ( y:ys, accum ) = rev ( ys, y:accum ) append– ([], ys) = ys in rev ( xs, [] ) append– (x:xs, ys) = x : append (xs, ys) • Reverse a list – reverse [] = [] 1 3 reverse– (x:xs) = (reverse xs) ++ [x] • Questions 2 2 2 2 3 3 1 3 1 1 – How efficient is reverse? – Can it be done with only one pass through list? List Comprehensions Datatype Declarations • Examples • Notation for constructing new lists from old: – data Color = Red | Yellow | Blue myData = [1,2,3,4,5,6,7] elements are Red, Yellow, Blue data Atom = Atom String | Number Int twiceData = [2 * x | x <- myData] -- [2,4,6,8,10,12,14] elements are Atom “A”, Atom “B”, …, Number 0, ... data List = Nil | Cons (Atom, List) twiceEvenData = [2 * x| x <- myData, x `mod` 2 == 0] -- [4,8,12] elements are Nil, Cons(Atom “A”, Nil), … Cons(Number 2, Cons(Atom(“Bill”), Nil)), ... • General form • Similar to “set comprehension” { x | x ∈ Odd ∧ x > 6 } – data <name> = <clause> | … | <clause> <clause> ::= <constructor> | <contructor> <type> – Type name and constructors must be Capitalized. Datatypes and Pattern Matching Case Expression Recursively defined data structure Datatype data Tree = Leaf Int | Node (Int, Tree, Tree) data Exp = Var Int | Const Int | Plus (Exp, Exp) 4 Case expression Node(4, Node(3, Leaf 1, Leaf 2), Node(5, Leaf 6, Leaf 7)) case e of 3 5 Var n -> … Const n -> … Plus(e1,e2) -> … 1 2 6 7 Recursive function sum (Leaf n) = n sum (Node(n,t1,t2)) = n + sum(t1) + sum(t2) Indentation matters in case statements in Haskell. 6 Laziness Evaluation by Cases Haskell is a