> -----Original Message-----
> From: Robert Stuart [mailto:[EMAIL PROTECTED]]
>
> Relational database theory derives from set theory. (EF Codd
> circa 1980)
>
> A tree structure may be viewed as a set of branches each
> containing a set
> branches and leaves.
> The root is simply a branch with no parent branch.
> A leaf is simply a branch with no child branch.
>
> Thus, we are mapping one set-of-sets onto another set-of-sets.
>
Each traversal of a relationship is an expensive operation. If you look at an XML document as a graph of nodes (ala Groves, etc.) the number of traversals to transform the data is very high. If each relationship is expensive, then XML document processing will be very expensive.
It would be better to store the XML as a blob, and extract XPath-style node identifiers for a searchable index (via SAX events as you store the blob).
Mike Dierken
"In theory, there is no difference between theory and practice. In practice there is."
