Stuff to do, Ideas, etc.

- Modify parser to create IR objects: (play with CUP actions)
   - Add types to non-terminals (see IR handout section 6)
   - Program is a TreeBlock of methods, each of which is a TreeBlock of
	TreeNodes (e.g. TreeIf, Instruction, etc)
   - Annotate TreeBlocks with SymbolTables.
        - Locals and parameters for methods
        - Field names for the Program
        - None for other blocks
   - Build SymbolTables at parse time (see IR handout section 6)
   - Flag some semantic errors during parsing?

- Modify Compiler.java to run Scanner and Parser, and then take new output
of Parser (the IR tree) and perform Semantic checking.

- Design decisions:
  - SymbolTable implementation (Hashtable? (in java.util I think))
     - getHierAnnotation??
  - Types?
      - Extend IRType. 
      - Annotate useful things (like Operands) with types
      - Include types in the symbol table, possibly as annotations, although
        not necessarily.
  - Procedures: Create a TreeProc class, as suggested on page 2 of IR handout?

- Semantic Rules
  - All the ones they list
  - What else is in the language spec document that we have to watch for also?

- Test cases
  - The given ones.
  - Our own, to test _every_ semantic rule

- Division of Labor? 
    - Symbol table implementation
    - Different CUP actions for different things (lots of stuff here)
    - Types
    - Other?