Re: A new approach to storing ordered hierarchical data in RDBs.

On Nov 22, 12:35 am, paul c <toledobythe..._at_oohay.ac> wrote:
> Lennart wrote:
> > Aloha Kakuikanu wrote:
> > [...]
>
> >>Please make sure you use terms correctly. What you have meant by
> >>"lineage" is "transitive closure". This article
> >>http://www.sqlteam.com/item.asp?ItemID=8866
> >>uses the term "lineage" in the sense of "materialized path".
>
> > I especially liked the phrase "Don't worry about the loop" :-)
>
> > /LennartI especially laughed at the phrase "XML handles hierarchical ... quite
> well". How the dickens does XML per se calculate transitive closure or
> depth from an adjency list or an adjaceny list from a tree? It seems to
> me that XML by itself only allows adhoc solutions that are just as
> non-declarative as various SQL versions. For that matter, the original
> relational model seems to requires similar adhoc additions.

Transitive reduction is hardly a new affair - Hasse Diagrams were doing the same job well before the advent of computer data models . However such diagrams are able to assume both transitivity and reflexivity for a partial order ing because they are /specific/ to the representation of such binary relations. Inferring that the relational model /needs/ to use an ad-hoc solution to achieve that goal misses the point that it is not confined to binary relations, and so as a generalized model, absolutely should not have specific rules for special cases of n-ary relations. This is a good thing, and naturally does not prevent one from designing another layer on top.

Of course, you are correct in your analysis of XML, which suffers from no such positive capabilities. Those that suggest XML is a good basis for a data model because it 'handles hierarchy natively', must be the same sort of people who'd choose a car based on how well it handles underwater. J.

> I don't
> have a solution, but it seems to me that all three are lacking and it is
> irresponsible, ignorant and cavalier to use the adverb "well" regarding
> accomplishing closure with any of them.
>
> p
Received on Wed Nov 22 2006 - 03:11:19 CET