Re: why hierarchy?

Joe Kesselman wrote:
> Reminder: XML is intended as an interchange syntax, and therefore was
> designed for maximum simplicity. By all means, if you want to manipulate
> the data, you should load it from the XML into whatever representation
> makes most sense -- which may be relational rather than hierarchical, or
> may be some other approach entirely, depending on the needs of the
> application. Don't confuse syntax, semantics, and implementation.
>
> On the other hand, IBM's new XML support in DB2 demonstrates that a
> direct representation of the XML infoset can indeed deliver good
> performance, if you're careful about coding it.
> ...

I think that should read "IF you're careful about coding it". I remember a Scientific American article by one or more of the XML originators where he admitted its redundancy and fobbed that off by suggesting compression techniques under the covers (ever wonder why people with older PC's are slow to uptake OpenOffice?). Golf patriots have a moral "play the ball where it lies". A data model moral might be "record facts/tags once".

> (I'm pointing this back to comp.databases.theory, since arguments about
> syntaxes other than XML really aren't relevant to comp.text.xml, and
> since you *can* represent relational data in XML if you want to do so.)

Well, at least it is admitted that XML is a mere syntax.

p Received on Wed Jul 26 2006 - 17:49:41 CEST