Re: LSP paradox

From: Cimode <cimode_at_hotmail.com>
Date: 19 Oct 2006 10:09:29 -0700
Message-ID: <1161277769.229891.267560_at_i3g2000cwc.googlegroups.com>

> I'm not sure what the other participants conclusion came to, but for me
> the best you can do is to have the same ground set for base class and
> subclass. The subclass would have more operations.

Anybody with a decent mathematical background would come to the conclusion that the set-subset *operator applyability cardinality* (forgive the barbaric expression) is a difficult one to demonstrate and should be supported by mathematical reasonning , NOT by historically made fuzzy computing concepts such as typing.

Beween superset D1 and derived subset d1 there are only 6 possible combinations to define *operator applyability cardinality* from D1 values to d1:

--> 1) Less values/ALL operations on D1 values apply on d1 values +
operations specific to d1 values.
--> 2) Less values/ALL operations on D1 values apply on d1 values
--> 3) Less values/SOME operations on D1 values apply on d1 values
--> 4) Less values/SOME operations on D1 values apply on d1 values +
operations specific to d1 values
--> 5) Less values/NO operations of applyable on D1 values apply on d1
values
--> 6) Less values/NO operations of applyable on D1 values apply on d1
values + operations specific to d1 values

It seems obvious that option 4) and 1) open most manipulative possibilities for a better RM implementations. All other combinations (2, 3, 5, 6) are restrictive. Received on Thu Oct 19 2006 - 19:09:29 CEST