Re: Clean Object Class Design -- Circle/Ellipse

"Marc Gluch" <marc.gluch_at_mindtap.com> wrote in message news:3b833bac.3938435951_at_news.grpvine1.tx.home.com...
> On Tue, 21 Aug 2001 19:42:25 GMT, "Richard MacDonald"
> <macdonaldrj_at_att.net> wrote:
>
> >Is there anything to be gained by using the axiom where input
> >and output types are the same, then viewing it from a generic
> >function's perspective where the type is Object, i.e.,
> >all-encompassing?
>
> I'm not sure I understand your question.
> Why are you reversing the order of construction
> (then viewing it...)?

Because I was starting from the "bottom" and going "up". You're starting at the "top" going "down". I think I just saw what you are getting at.

> You start with total Chaos.

Let there be Object.

> You "bring some order to the chaos" by stating first axiom.

Let there be some function of Object which produces Object.

> With addition of each new axiom, you build up (actually down) new
> subtheory. The type of each operation completely depends on the
> axiom(s) defining that operation.

Let there be some function of Object A which produces Object B.

1. If B is a subtype of A, you're going "down". No problem. Very clear.
2. If B is the same type of A, no problem.
3. If A is a subtype of B, where does the function go?
4. If neither A or B is a subtype of each other, where does the function go?

I should think 3 is a sub-problem of the more general problem 4, but that is an assumption right now.

3a) The function is defined in A?
3b) The function is defined in B?

4a) The function is defined in A?
4b) The function is defined in B?
4c) The function is defined in the LCS (lowest-common supertype) of A and B.

I read you as saying (3b).
How about 4?

(Now that I think I clicked, I can probably go back and figure out your terminology, but my language might be easier for others to read?

I like the way your axioms are constraining the space. As an n-dimensional applied math guy, I like the picture. Received on Sat Aug 25 2001 - 01:44:28 CEST