home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!edcastle!dcs.ed.ac.uk!rjp
- From: rjp@dcs.ed.ac.uk (Rob Pooley)
- Newsgroups: comp.object
- Subject: Re: Object-Oriented Methodologies - Class Specifications
- Message-ID: <43008@skye.dcs.ed.ac.uk>
- Date: 2 Sep 92 08:58:20 GMT
- References: <715276480.1.p00058@mail.psi.net> <1992Sep1.220559.10346@m.cs.uiuc.edu>
- Sender: nnews@dcs.ed.ac.uk
- Organization: Department of Computer Science, University of Edinburgh
- Lines: 22
-
- The danger of navel inspection is well known. Whilst I agree that the definition of a
- circle is fascinating it doesn't really have any direct connection to designing a
- class library.
-
- If I ask my children to define a circle, they will draw one. The mathematicians may
- have fancy definitions, but my kids don't know that. Since their "definition" is
- older than the mathematicians' which is right?
-
- You define any class in the context of its use. Any methodology that doesn't know
- this is going to take you down the same old blind alleys as past "problems" with
- OO techniques. I make draw into a method of circle if that helps me build my package
- in a way that is easy to understand and to use. It may invoke a draw method in the
- screen or printer or strean object to which it is currently connected. T%his is why
- we need virtual and the ability to tell an object about its environment (Parameters
- as we used to call them).
-
- Keep it simple and remember what you are trying to do. Don't fit the problem to the
- method, but fit the method to the problem. You could both be right (or wrong)
- depending on the context.
-
- Rob Pooley
- Computer Science, Edinburgh University
-