Cart
[Home] [By Thread] [By Date] [Recent Entries]
Mike Champion wrote:
>
>
>> Object models are
>> *local*. Syntax is *global*.
>
> The success of XSLT, XPath, and the (probable) success of at least the
> hard core of XQuery is a strong counter-example to this assertion.
It's much simpler than this, I think. There are two kinds of things you
do with XML. First, you use it to interchange data among disparate
systems. Second, you use it as a basis for processing once you've
received it.
The first (interchange) scenario has no requirement whatsoever for a
shared data model: there are a million existence proofs on the ground of
interoperating applications who defined the syntax, got on with the job,
and are now in production.
The second (processing) scenario obviously presupposes a data model,
ranging from the moronically-simple SAX event stream to a detailed
object model with fully worked out thread-safety and concurrency and so
on. The degree to which object models can be shared from one
programming language to another, or from one OS to another, varies
wildly. People are claiming thath XPath model works well for a wide
variety of processing scenarios. I'm inclined to believe this based on
what I've seen so far.
Having said that, I think the interchange usages of XML will for the
foreseeable future outnumber and outweigh the processing uses.
--
Cheers, Tim Bray
(ongoing fragmented essay: http://www.tbray.org/ongoing/)
|
