CSE 341
Lecture 15
introduction to Scheme
slides created by Marty Stepp
http://www.cs.washington.edu/341/
Looking back: Language timeline
category 1960s 1970s 1980s 1990s 2000s
scientific Fortran Matlab
business Cobol DBMSes SQL VB
functional Lisp ML, Scheme Erlang Haskell F#
imperative/ Algol Pascal, C, Ada, C++ Java C#
procedural Smalltalk
scripting BASIC Perl Python,
Ruby, PHP,
JavaScript
logical Prolog CLP(R)
2
History of LISP
• LISP ("List Processing"): The first functional language.
made: 1958 by John McCarthy, MIT (Turing Award winner)
– godfather of AI (coined the term "AI")
developed as a math notation for proofs about programs
pioneered idea of a program as a collection of functions
became language of choice for AI programming
• Fortran (procedural, 1957), LISP (functional, 1958)
languages created at roughly the same time
battled for dominance of coder mindshare
Fortran "won" because LISP was slow, less conventional
3
John McCarthy, creator of LISP
4
LISP key features
• a functional, dynamically typed, type-safe, language
anonymous functions, closures, no return statement, etc.
less compile-time checking (run-time checking instead)
accepts more programs that ML would reject
• fully parenthesized syntax ("s-expressions")
Example:
(factorial (+ 2 3))
• everything is a list in LISP (even language syntax)
allows us to manipulate code as data (powerful)
first LISP compiler was written in LISP
5
LISP advanced features
• LISP was extremely advanced for its day (and remains so):
recursive, first-class functions ("procedures")
dynamic typing
powerful macro system
ability to extend the language syntax, create dialects
programs as data
garbage collection
continuations: capturing a program in mid-execution
• It took other languages 20-30 years to get these features.
6
LISP "today"
• current dialects of LISP in use:
Common LISP (1984) - unified many older dialects
Scheme (1975) - minimalist dialect w/ procedural features
Clojure (2007) - LISP dialect that runs on Java JVM
• well-known software written in LISP:
Netscape Navigator, v1-3
Emacs text editor
movies (Final Fantasy), games (Jak and Dexter)
web sites, e.g. reddit
Paul Graham (tech essayist, Hackers and Painters)
7
Scheme
• Scheme: Popular dialect of LISP.
made in 1975 by Guy Steele, Gerald Sussman of MIT
Abelson and Sussman's influential textbook:
– Structure and Interpretation of Computer Programs (SICP)
http://mitpress.mit.edu/sicp/
• innovative differences from other LISP dialects
minimalist design (50 page spec), derived from λ-calculus
the first LISP to use lexical scoping and block structure
lang. spec forces implementers to optimize tail recursion
lazy evaluation: values are computed only as needed
first-class continuations (captures of computation state)
8
TeachScheme!
• 1995 movement by Matthias Felleisen of Rice's PLT group
goal: create pedagogic materials for students and teachers
to educate them about programming and Scheme
push for use of Scheme and functional langs. in intro CS
radical yahoos who take themselves too seriously :-)
• major TeachScheme! developments
DrScheme editor, for use in education
How to Design Programs, influential Scheme intro textbook
http://www.teach-scheme.org/
http://www.htdp.org/
9
DrScheme
• DrScheme: an educational editor for Scheme programs
built-in interpreter window
– Alt+P, Alt+N = history
syntax highlighting
graphical debugger
multiple "language levels"
– (set ours to "Pretty Big")
• similar to DrJava editor for Java programs
(you can also use a text editor and command-line Scheme)
10
Scheme data types
• numbers
integers: 42 -15
rational numbers: 1/3 -3/5
real numbers: 3.14 .75 2.1e6
complex/imaginary: 3+2i 0+4i
• text
strings: "\"Hello\", I said!"
characters: #\X #\q
• boolean logic: #t #f
• lists and pairs: (a b c) '(1 2 3) (a . b)
• symbols: x hello R2D2 u+me
11
Basic arithmetic procedures
(procedure arg1 arg2 ... argN)
• in Scheme, almost every non-atomic value is a procedure
even basic arithmetic must be performed in () prefix form
• Examples:
(+ 2 3) → 5 ; 2 + 3
(- 9 (+ 3 4)) → 2 ; 9 - (3 + 4)
(* 6 -7) → -42 ; 6 * -7
(/ 32 6) → 16/3 ; 32/6 (rational)
(/ 32.0 6) → 5.333... ; real number
(- (/ 32 6) (/ 1 3)) → 5 ; 32/6 - 1/3 (int)
12
More arithmetic procedures
+ - *
quotient remainder modulo
max min abs
numerator denominator gcd
lcm floor ceiling
truncate round rationalize
expt
• Java's int / and % are quotient and modulo
remainder is like modulo but does negatives differently
• expt is exponentiation (pow)
13
Defining variables
(define name expression)
• Examples:
(define x 3) ; int x = 5;
(define y (+ 2 x)) ; int y = 2 + x;
(define z (max y 7 3)) ; int z = Math.max..
• Unlike ML, in Scheme all top-level bindings are mutable!
(set! name expression)
(set! x 5)
– (Legal, but changing bound values is discouraged. Bad style.)
14
Procedures (functions)
(define (name param1 param2 ... paramN)
(expression))
• defines a procedure that accepts the given parameters
and uses them to evaluate/return the given expression
> (define (square x) (* x x))
> (square 7)
49
in Scheme, all procedures are in curried form
15
Basic logic
• #t, #f ; atoms for true/false
• <, <=, >, >=, = operators (as procedures); equal?
(< 3 7) ; 3 < 7
(>= 10 (* 2 x)) ; 10 >= 2 * x
• and, or, not (also procedure-like; accept >=2 args) *
> (or (not (< 3 7)) (>= 10 5) (= 9 6))
#t
(technically and/or are not procedures because they don't always
evaluate all of their arguments)
16
The if expression
(if test trueExpr falseExpr)
• Examples:
> (define x 10)
> (if (< x 3) 10 25)
25
> (if (> x 6) (* 2 4) (+ 1 2))
8
> (if (> 0 x) 42 (if (< x 100) 999 777)) ; nested if
999
17
The cond expression
(cond (test1 expr1) (test2 expr2)
... (testN exprN))
• set of tests to try in order until one passes (nested if/else)
> (cond ((< x 0) "negative")
((= x 0) "zero")
((> x 0) "positive"))
"positive"
• parentheses can be []; optional else clause at end:
> (cond [(< x 0) "negative"]
[(= x 0) "zero"]
[else "positive"])
"positive"
18
Testing for equality
• (eq? expr1 expr2) ; reference/ptr comparison
• (eqv? expr1 expr2) ; compares values/numbers
• (= expr1 expr2) ; like eqv; numbers only
• (equal? expr1 expr2) ; deep equality test
(eq? 2.0 2.0) is #f, but
(= 2.0 2.0) and (eqv? 2.0 2.0) are #t
(eqv? '(1 2 3) '(1 2 3)) is #f, but
(equal? '(1 2 3) '(1 2 3)) is #t
Scheme separates these because of different speed/cost
19
Scheme exercise
• Define a procedure factorial that accepts an integer
parameter n and computes n!, or 1*2*3*...*(n-1)*n .
(factorial 5) should evaluate to 5*4*3*2*1, or 120
• solution:
(define (factorial n)
(if (= n 0)
1
(* n (factorial (- n 1)))))
20
List of Scheme keywords
=> do or
and else quasiquote
begin if quote
case lambda set!
cond let unquote
define let* unquote-splicing
delay letrec
• Scheme is a small language; it has few reserved words
21
