home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cimshop!davidm
- From: davidm@consilium.com (David S. Masterson)
- Newsgroups: comp.object
- Subject: Re: Object-Oriented Methodologies - Class Specifications
- Message-ID: <DAVIDM.92Sep7024740@consilium.com>
- Date: 7 Sep 92 09:47:40 GMT
- References: <graham.715679543@galois> <1992Sep6.003055.2236@tfs.com>
- <graham.715829976@galois> <1992Sep7.034245.602@tfs.com>
- <graham.715854518@galois>
- Sender: root@cimshop.UUCP
- Distribution: comp
- Organization: Consilium Inc., Mountain View, California
- Lines: 47
- In-reply-to: graham@maths.su.oz.au's message of 7 Sep 92 08:28:38 GMT
- X-Posting-Software: GNUS 3.13 [ NNTP-based News Reader for GNU Emacs ]
-
- >>>>> On 7 Sep 92 08:28:38 GMT, graham@maths.su.oz.au (Graham Matthews) said:
-
- > eric@tfs.com (Eric Smith) writes:
-
- >>The trick is to see that relationships are themselves objects. For example,
- >>the ways a line and a circle can relate to each other are the line-circle
- >>relationship, which is a subclass of the two-geometric-figure relationship.
- >>Intersections are a feature of geometric figures, and the intersections of a
- >>line and a circle are a feature of the line-circle-relationship subclass of
- >>the two-geometric-figure relationship.
-
- > Three points in reply:
-
- > a) I find it very unnatural to think of the relationships between objects as
- > objects. This is why we have two names for the two different ideas.
-
- In the Smalltalk world, everything is an object. Put another way, when you
- can apply attributes to a relationship (in this case, the point where two
- figures intersect), then maybe the relationship is also an entity in its own
- right. The classic question from E/R modelling is "what is a marriage -- a
- relationship between two people or an event that occurred at a specific time
- and place?"
-
- > b) you assume that the relationships between objects can be modelled using
- > your inheritance structure, or more generally that the mechanism for
- > classifying the types of objects is to be the same mechanism for handling
- > relationships between types. I see no reason to assume that this assumption
- > is true.
-
- Why not? It is one model of the problem.
-
- > c) if you have a lot of relationships your scheme gets pretty clumsy.
- > Moreover many of these relationships are not naturally specified in an
- > inheritance framework.
-
- Yes, keeping track of all those potential relationships between objects won't
- be pretty, but that's what E/R modelling is for. When relating two geometric
- objects, for instance, there would be a lot of double dispatching to try and
- figure out what the exact relationship is.
- --
- ====================================================================
- David Masterson Consilium, Inc.
- (415) 691-6311 640 Clyde Ct.
- davidm@consilium.com Mtn. View, CA 94043
- ====================================================================
- "Politics is for the moment. An equation is for eternity."
- -- Albert Einstein
-