Re: Nearest Common Ancestor Report (XDb1's $1000 Challenge)

> > color isa thing. brown isa color. person isa thing. john isa person.
> > john's color is brown. mary isa person. mary's color is brown.
> >
> > Entering the equivalent of the above XDb1 script in RM Sol#1 or #2 results
> > in redundant data because "brown" is stored once in T_thing and twice in
> > T_attributes_char. Changing one of them and not the others, would
> > essentially corrupt the data.
>
> if we have stored a person with surname "Brown" and he tells us that
> actually it is spelt "Browne" then you will avoid an update anomaly
> by changing the word in one place, and now you will have
> "my car is Browne" too!

You are comparing two different case. If the first case, the three redundant "brown"s are the same name for the same thing, a color. There was no person named "brown".

In the case you presented, the name "brown" was that of two different things, a person and a color. When the person's name is changed from "brown" to "browne" the proper modification is to unlink the person from "brown" and relink to a new name "browne". XDb1's GUI does exactly this when person named "brown" is renamed to "browne". In the case where "brooown" is later determined to be the correct name for both the person and color, one only needs to update the name "brown" itself to "brooown" and now all things (person and color) retrieve their correct name. Received on Thu Jun 10 2004 - 18:30:32 CEST