CS 5010: Guided Practice 4.4 — Solution

The question was:

Which of the following are sensible data definitions, according to Lesson 4.4? Why or why not? Write a template for those that are sensible.

    A Foo is one of
    -- empty
    -- (list Number Foo)
    
    {how is this different from a a ListOfNumber}?
    
    Answer: this is sensible-- here is the template
    
    (define (foo-fn x)
      (cond
        [(empty? x) ...]
        [else (.. (first x)
                  (foo1fn (second x)))]))
    
    We distinguish the two cases with empty?, and we get the fields of x
    with (first x) and (second x).
    
    Note that we recur on (second x) instead of (rest x).
    
    A Blaster is one of
    -- empty
    -- (cons Nat (cons Nat Blaster))
    
    Answer: this is sensible-- we distinguish the two cases with empty?,
    and we get the fields with (first x), (second x), and (rest (rest x)).
    
    The template is
    
    (define (blaster-fn x)
      (cond
        [(empty? x) ...]
        [else (... (first x)
                   (second x)
                   (blaster-fn (rest (rest x))))]))
    
    Note that a Blaster always has an even number of elements.
    
    While this is data definition is sensible, it is not a very good data
    design.  Almost certainly it is preferable to write
    
    A BetterBlaster is a ListOfBlasterData
    
    where BlasterData is a struct containing the two Nats, which you
    define elsewhere.
    
    A Gargle is one of
    -- (list Nat)
    -- (cons Nat (cons Nat Gargle))
    
    Answer: this is also sensible.  We distinguish the two cases with
    (empty? (rest x)), and we get the fields  (first x), (second x), and
    (rest (rest x)).
    
    The template is
    
    (define (gargle-fn x)
      (cond
        [(empty? (rest x)) (... (first x))]
        [else (... (first x)
                   (second x)
                   (gargle-fn (rest (rest x))))]))
    
    Note that a Gargle always has an odd number of elements.  Like a
    Blaster, this is sensible, but it is not a good design.  If you ever
    write something like this, go back and think about your problem harder.
    
    A GargleBlaster is one of
    -- a Gargle
    -- a Blaster
    
    This data definition is sensible, but it is not very good.  To
    distinguish the two cases, you'd have to ask (even? (length x)), which
    isn't really a "simple" predicate.  Worse yet, your partner might change
    the definition of either Gargle to something like
    
    A Gargle is one of
    -- (list Nat)
    -- (cons Boolean Gargle)
    -- (cons Nat (cons Nat Gargle))
    
    Now a Gargle might have either an even or odd number of elements, and
    (even? (length x)) won't distinguish the two cases.
    
    If you ever write something like this, you probably wrap your Gargle
    or your Blaster in a struct, maybe something like this:
    A GarbleBlaster is one of
    -- (make-gargle-gb Gargle)
    -- (make-blaster-gb Blaster)
    
    (define-struct gargle-gb (data))
    (define-struct blaster-gb (data))
    
    The template for this data definition is
    
    (define (garbleblaster-fn gb)
      (cond
        [(gargle-gb? gb) (gargle-fn (gargle-gb-data gb))]
        [(blaster-gb? gb) (baster-fn (blaster-gb-data gb))]))
  

And this is guaranteed to work now no matter how your partner changes the definition of a Gargle.