Summary of Class Presentation for September 9, 1999

	Topics:  Review and Evaluation of What We have Done Thus Far

	In the first two weeks of this course we have covered the topics of
what makes a system object-oriented and two design methods for object-
oriented systems.  A system is object-oriented if the system consists of 
objects where each object has the following properties:
	1. may contain private, static data;
	2.  may contain a set of operations that use the private data;
	3.  encapsulates the private data so that it cannot be used 
		outside of this object;
	4.  provides an interface that consists of a set of messages 
		accepted by this object (each message corresponds to 
		one publically available operation implemented within
		the object);
	5.  communicates with other objects solely by sending them 
		messages and receiving values as the result of those 
		messages.  A message differs from a function call in 
		that the message specifies a receiver as well as the 
		actual message.
Further, most object-oriented systems have classes that provide a model or
template for the data and operations to be available in the objects that
are instances of each class.  The class itself might have operations and
data of its own as well.  We will see later that there are three 
important features of object-orientation: dynamic polymorphism, upcasting,
and downcasting that provide the power of the object-oriented approach.

	Classes have two roles:
1.  serve as a template, framework, or model for a set of instances(objects
),  The class provides a type for the objects;
2.  provide data and operations that involve the entire set of objects as
a whole (for example, keeping track of how many objects in this type 
exist at any time during execution, searching among the objects of the
type for a particular one, moving an object of this type dynamically to 
another computer over a network).

	Objects have two roles:
1.  provide an interface that specifies the services this object provides
for other objects and classes;
2.  implement the interface in an encapsulated way.

	The two object-oriented design methods presented were the Abbott
noun method, and a data collections-centered method.  The Abbott method
is representative of those design methods that emphasize the intended 
behavior of the product.  The data collections-centered method is 
representative of those design methods that emphasize the modeling aspect
of object-orientation.  The two methods actually follow approximately the 
same major steps:
- Understand the problem:
	Abbott: read and understand the requirements
	DCC:  determine inputs, outputs, and the relationships between 
		them

- Develop candidate set of classes:
	Abbott: nouns in the requirements description
	DCC:  data collections and standalone scalar variables

-  Winnow the candidate set to form a smaller set of top-level classes:
	Abbott:  heuristics
	DCC: wlakthrough scenarios (use cases)

- Evaluate the architectural design:
	Abbott:  cohesion, coupling, simplicity, clarity
	DCC:  simplicity, clarity, cohesion, coupling.