CSC302 2006S, Class 42: Wrapup

Admin:
* Sorry about last night.

Overview:
* Grading.
* About the final.
* About the midsem.
* Wrapup.
* Evaluation forms.

Final Topics:
* Languages
  * Prolog: Probably small program ; Possibly hybrid
  * Haskell: Focus on laziness or patterns or HOP; Possibly hybrid
  * FP: Concepts
  * Ruby
  * SQL
* Concepts
  * Denotational semantics
  * Polymorphism
  * Types
  * Future of languages
* Closed book, closed computer
  * One page of notes, 8.5 x 11 inches (or A4), double-sided, hand-written

Midsem
* Consider
  v1 : f1 = short -> short
  v2 : f2 = short -> int
  v3 : f3 = int -> int
* Claim: In a general form of equiv, we might allow
  vi = vj  where fi and fj are functional types
   if the domain of fj is a subset of the domain of fi
   and the range of fj is a subset of the range of fi
   BAD IDEA TO PERMIT
   suppose short = 1..10
   suppose int = 1..100

   v3(50) OK
   v1(50) BOOM.  OUTSIDE OF DOMAIN
   v3 = v1

   THE CLAIM IS INCORRECT
Problem 1:
  Assume E => I = E
* First important consequence is that expressions can change the env

meaningExp :: Exp Env -> N
  WHOOPS NEED TO RETURN A NEW ENVT, TOO

ONE POSSIBILITY
meaningExp :: Exp Env -> (N x Env)

ANOTHER POSSIBILITY
meaningExp :: Exp Env ECont -> ...

NEED TO REWRITE EVERYTHING THAT USES meaningExp TO USE THE POTENTIALLY-CHANGED ENVT

NEED TO REWRITE E1 + E2

Problem 2:

(define cachedcont 0)
(define suspend
  (lambda ()
    (let/cc k
      (let ((resume cachedcont))
        (set! cachedcont k)
        (resume)))))