[
Lists Home |
Date Index |
Thread Index
]
On Friday 24 October 2003 11:34, Simon St.Laurent wrote:
> tyler@waterken.com (Tyler Close) writes:
> >That's what a data model is. A data model tells you how you can
> >expect data to be structured.
>
> Perhaps the problem is that I make a distinction between "data" and "XML
> document". I don't generally give a damn about sharing structures
> beyond the document itself. When I do, I try to keep those structures
> as minimal as possible.
But the document itself structures data. The schema for the
document is a description of how the data is structured. You are
using a data model. The schema defines the data model.
When you write code to fetch a string from an XML document, how do
you know where in the XML document to look for the string? The
XML specification doesn't tell you where to find the string. You
must have some kind of data model that tells you where to look for
the string.
> >Perhaps we need an explicit definition of what a data model is.
> >Could you provide yours?
>
> A data model can range from a blessedly custom-fit outfit tailored
> specifically to a particular need, or a hair-shirt straitjacket meant to
> be shared by as many people as possible.
>
> My problem is not with the notion of a data model per se - I use them in
> my own code all the time. My problem is with the notion of shared data
> models. Sharing syntax, fine. Sharing data models? That costs a lot,
> in various ways.
>
> Shared data models are poison.
If that's your best shot at a definition, you don't have a clear
notion of what a data model is. Maybe that's why you call it
poison.
Tyler
|
