Re: Clean Object Class Design -- Circle/Ellipse

[posted and mailed]

slong3_at_mediaone.net (Steve Long) wrote in <JNgg7.2200$L8.32631_at_typhoon.jacksonville.mediaone.net>:

>having a master of science in theoretical mathematics, perhaps some
>would agree that i may provide an informed perspective on the topics at
>hand.
>
>first, wrt two dimensional geometries, there are really only two kinds:
>closed curves and open curves. any closed curve can be mapped into any
>other closed curves by applying some function(s); ie, all closed curves
>are isomorphic. similarly, all open curves are isomorphic. it makes no
>difference what you choose to label your closed curve... it is just an
>alternative representation of any other closed curve. there are well
>known proofs of these statements.

I think you're talking about topological equivalence. Topologically, all closed *simple* curves are equivalent. If you have selfintersections, things get more complicated. Also, there's no reason to assume you can usually lump all simple closed curves together. The difference between squares and circles can matter, as anyone who's tried to fit a square peg into a round hole can tell you;-)

>second, wrt classes, whether or not one subclasses one from the other
>just depends. it may or may not be useful. this is a design decision
>to be made by a "hopefully" informed application architect or developer.

Agreed.

>unless there is a clear understanding of the scope and nature of the
>project, one has a hard time making a determination as to which approach
>is optimal. one even must take into account the performance
>characteristics of the programming language and platforms. for example,
>while Ada is very flexible wrt generic classes and generic instantations
>as well as subclassing, etc, it is a resource hog, as is java. the
>dependencies in Ada results in minor chagnes in the original base class
>or significantly distant ancestors a major effort to recomplie (the
>entire application).

Good points.

Regards,

Martijn Received on Tue Aug 21 2001 - 18:20:57 CEST