Additional questions:

 - What is the scope of the bindings of the accessors?  Can I write

   (define-type point (previous)
     (fields (x (point-x)) (point-x previous))
             (y (point-y)) (point-y previous))))

 - Same question for the constructor and predicate.

 - What is the scope of the binding of the type itself?  Can I write

   (define-type foo ()
     (fields (x mutable) (......(type-descriptor foo).......)))

- Can I use (parent <parent name> ...) in the syntactic layer to extend
  a record type that has been defined procedurally?
  Again, if so, would seem to imply that <record name>s are implicitly
  quoted symbols, which brings up the question of how they are bound
  in e.g. local define-type's.

- Related to previous:  Do <record name>s belong to a separate namespace
  from ordinary identifiers?

- Does TYPE-DESCRIPTOR need to be syntax, or can it be a procedure?

 Cheers
 Andre