| |
| | Computer Science 68, Notes 01-27 (Site not responding. Last check: 2007-10-21) |
 | | In theory, laziness is "optimal" when compared to eager or evaluate-by-need; in practice, there's some extra overhead to handle the memoization, so real-world eager languages may still be a bit faster than real-world lazy languages on most programs. |
 | | The problem with the "exists(t x) e" expression, and the evaluation rule given above, is that it is hard to implement: the rule suggests that an interpreter has to guess a value v, substitute v in for x in the expression e, and then check to see that the result evaluates to true. |
 | | Almost all the evaluation rules and type checking rules for jLogic are exactly the same as the rules for j2 (except that you have to write "p" in many places where an "e" occurs in j2, which I kept forgetting in class). |
| www.cs.dartmouth.edu /~cs68/04w/notes-01-28.html (2410 words) |
|