Re: why hierarchy?
From: paul c <toledobythesea_at_oohay.ac>
Date: Wed, 26 Jul 2006 15:13:44 GMT
Message-ID: <IELxg.246187$IK3.209770_at_pd7tw1no>
>
> Could someone demonstrate the advantages of XML or RM at representing
> the simple example shown at www.dbfordummies.com/example/ex007.asp
> This example represents John and Mary in two hierarchies (without
> redundancy). In the first hierarchy, John and Mary are children of Adam
> and Eve who are children of God. In the second hierarchy, John and Mary
> are part of Mars and Venus respectively which are part of the Universe.
>
> To verify which method's underlying data model is more general, one can
> observe the impact on existing schema/data/query when meeting future
> unknown data requirements (to be proposed by someone else). Would
> someone be willing to engage in this trivial example?
>
Date: Wed, 26 Jul 2006 15:13:44 GMT
Message-ID: <IELxg.246187$IK3.209770_at_pd7tw1no>
Neo wrote:
>>> I spoke in a recent thread of my distaste for XML from a theoretical standpoint ... I find this return of focus to hierarchical models since 1995 extremely frustrating. >> Addressing the idiocy above, the complexity and richness of the data are inherent to the data remaining with it no matter the representation. Increasing the complexity of the representing structures only increases complexity without adding any 'richness' at all. 'Richness' comes from the available operations for manipulating data represented in those structures, which are greatly reduced as the structural complexity increases.
>
> Could someone demonstrate the advantages of XML or RM at representing
> the simple example shown at www.dbfordummies.com/example/ex007.asp
> This example represents John and Mary in two hierarchies (without
> redundancy). In the first hierarchy, John and Mary are children of Adam
> and Eve who are children of God. In the second hierarchy, John and Mary
> are part of Mars and Venus respectively which are part of the Universe.
>
> To verify which method's underlying data model is more general, one can
> observe the impact on existing schema/data/query when meeting future
> unknown data requirements (to be proposed by someone else). Would
> someone be willing to engage in this trivial example?
>
It does seem trivial. One hiearchy is talking about ancestors of persons, the other about origins of persons. The hierarchies record the persons twice. Relations would record them once and record ancestors once and record origins once. Many problems with hierarchies - here's one: if a user is aware of only one hierarchy and discovers that there is no such person as John and 'deletes' him, the other hierarchy is now wrong, as is the DB.
p Received on Wed Jul 26 2006 - 17:13:44 CEST