[
Lists Home |
Date Index |
Thread Index
]
At 03:36 PM 5/23/2003, Simon wrote:
>Objects are great if everyone agrees on the objects. In my experience,
>no one really wants to go that far, and information representation is a
>useful middle ground.
In addition to the other points made, it's worth noting that although many
XML data models in use are trees, there is quite a bit of variation between
one and another, even among those that are specified publicly -- compare
the various flavors of DOM to the XPath 1 model, then look at XPath 2, etc.
Standardizing on the syntax has allowed us to get the benefit of network
effects among tools that handle that syntax. Some may think that implied in
this syntax, waiting to get out, is a Universal Data Model that really
ought to fit every processing scenario (or any processing scenario of
interest) -- but that's easier to believe before we've actually tried to
define what that model is. (Are entities in it? How about CDATA sections?
characters?)
If XML had been defined on an explicit data model five years ago, it'd
probably be quite unsuitable for many of the things people are now doing
with it.
(Not that I believe that specifying a model is necessarily a bad thing: I
don't.)
Cheers,
Wendell
___&&__&_&___&_&__&&&__&_&__&__&&____&&_&___&__&_&&_____&__&__&&_____&_&&_
"Thus I make my own use of the telegraph, without consulting
the directors, like the sparrows, which I perceive use it
extensively for a perch." -- Thoreau
|
