Object-Oriented Search
----------------------

Goal: Design a class hierarchy for general-purpose breadth-first,
      depth-first, uniform-cost and best-first search.

Search Engine (solve)
 1. Put start node on OPEN
 2. Get first node on OPEN.  If none, Fail.
 3. Remove first node N from OPEN and put on CLOSED.
    If n is a goal, then return n.
 4. Expand N to obtain successors.
 5. Put new successors on OPEN.  Goto 2.  Insertion on OPEN depends
    on existence and heuristic value.

Example: 8-puzzle
- state is a 3x3 array

OPEN and CLOSED lists
- doubly-linked list class (LIST_)
- sorted doubly-linked list class (SORTEDLIST_ : LIST_)
- list iterator class (LIST_ITERATOR_)

NODE representation
- state description (depends on problem)
- state equality (test uniqueness and goal condition)
- apply operators (do_operator(int))
- display

Search Problem
- problem initialization (constructor)
  - number of operators
  - initial and goal states
- OPEN and CLOSED lists
- adding to lists (add(NODE_ *))
- search engine (solve)

Tree Search vs. Graph Search
- tree assumes all states unique
- graph assumes a state arrived at thru multiple paths
  - keeps best path
- affects whether children added to OPEN