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

From: Nick Landsberg <hukolau_at_worldnet.att.net>
Date: Mon, 21 Jun 2004 00:12:15 GMT
Message-ID: <zJpBc.105384$Gx4.21453@bgtnsc04-news.ops.worldnet.att.net>

Neo wrote:

>>>Although brown appears three times in the script, XDb1 normalizes the
>>>second two by making them references to the first one. There is only
>>>one brown in XDb1's db. Using RM#1 or #2, brown is stored three
>>>separate times.
>>
>>Would that mean that if John's name is brown, it would be stored in
>>the same place as the color brown?

>
>
> No, both the person named brown and the color named brown have
> references to a single name brown. In XDb1, a name (ie 'brown') is a
> thing, a person (named brown) is a thing and a color (named brown) is
> a thing. A thing be related to 0 to many names. A name can be related
> to 0 to many things.
>
>

>>And would that mean that if the UI designer wanted to change all browns 
>>to umber as part of a UI redesign, that it would change John's name?

>
>
> The person named brown doesn't store the name brown, it refers to it.
> The color named brown doesn't store the name brown, it refers to it.
> Thus if the name brown is changed to umber, all things named brown
> appear to be renamed to umber. When the GUI displays brown the person
> and brown the color, their names are being retrieved by referencing
> the same name.

So, are you saying that if Christopher Browne's name was misspelled in the original input as "Brown", and then corrected somehow, Alice Brown's name would become Alice Browne (note spelling) in any reports generated?

What is the mechanism for correcting such (normally encountered) situations?

(Hint: In a "normal" database structure, it would mean replacing one attribute in C. Browne's record. As I understand your scheme, if you did it right, it would mean searching the data-store for any "things" which contained the string "Browne", if found, creating a reference to that in C. Browne's *new* record, and replacing the record. If not found, creating a thing with the content "Browne", and then replacing C. Browne's record with the new reference. Seems like an awful lot of work to change a misspelling.)

To take this to the point of silliness, why doesn't the string "brown" refer to the characters 'b', 'r', 'o', 'w', 'n' as 5 different references which could each be stored in the data-store just once. After all 'b' is a thing, isn't it?

/Reductio ad absurdum./

-- 
"It is impossible to make anything foolproof
because fools are so ingenious"
  - A. Bloch

Received on Sun Jun 20 2004 - 19:12:15 CDT