Re: [xml-dev] Caught napping!

[Ronald Bourret <rpbourret@rpbourret.com>]

Dan Weinreb wrote:
> 
>    From: Ronald Bourret <rpbourret@rpbourret.com>
>
>    It is. But if you fully normalize your data and then XLink everything
>    together, you've just built a relational database, so why not just use
>    one to start with?
> 
> (1) You don't have to flatten everything completely in order to avoid
> redundancy (see my earlier mail).
> 
> (2) Maybe you want to be using XML documents for various reasons.
> Consider the ebXML Collaboration Partner Agreement: it's a document
> that spells out a business agreement between two partners.

I agree with both points, and I admit to being a bit facetious in my
reply. What I wanted to point out was that many people think about
native XML databases in terms of the data currently stored in relational
databases. This leads to questions of normalization and then to creating
a relational database in a native XML database. It makes more sense to
use a native XML database when a relational database won't do.

>  We want it
> to be a document, so that we can email it to each other, so that we
> can both look at it in a text editor, and so on.

But this doesn't imply it has to be stored in a native XML database as
opposed to a relational database.

>  If the information
> were stored in a relational database, it would not be as convenient
> to transmit the data from one place to another nor to look at it.

I think this depends entirely on the structure of the document. For
purely document-centric documents, this is true and for purely
data-centric documents, it probably isn't. There's a nice, big gray area
in the middle :)

> (Nevertheless, once you have a CPA document, you might want to
> store it in a database where it can be kept persistent, queried,
> and so on.)

Agreed. The history aspects of native XML databases were ones I hadn't
considered before this conversation.

-- Ron