4526P@NAVPGS.BITNET ("LT Scott A. Norton, USN") (12/17/86)
Thank you, Larry, for the clear explaination of OOP. I would like to add a few comments to address the design phase that preceeds OOP. These comments are based on the work I've read of Dr. David Parnas. To enable OOP to "tend to produce programs that need little change or can be easily changed," the objects that are defined should encapsulate those aspects of the project likely to change. Parnas calls these details the secrets of the module that defines the objects. Then, when a requirement changes, the program change is isolated to that module. For object-oriented programming to be totally effective, the design of the system must explicitly consider what design decisions are likely to change, and use that as the criterion to define objects. As Parnas found in his study of the A-7 aircraft's tactical program, this design requires significant amounts of research, and should not be a decision that the programmer makes sitting at a terminal. The designers must invest extra time before coding starts. LT Scott A. Norton, USN Naval Postgraduate School Monterey, CA 93943-5018 <4526P@NAVPGS.BITNET>