[
Lists Home |
Date Index |
Thread Index
]
On Fri, 2002-05-10 at 19:40, Ronald Bourret wrote:
> I've always assumed people would transmit the XML instance and a schema
> and that the processor would generate the PSVI (or those parts of it
> relevant to the application).
That's where I started out as well. (And I should add storage to the
transmission story.)
Then I started dealing with a lot more SOAP, where xsi:type is awfully
popular with or without a schema, and started looking through the XQuery
drafts. Thinking more and more about how tightly developers seem to
want their information wedded with its type and how little patience I'm
finding for actually writing code that walks through that process took
me another few steps down the path.
I'm not sure what flipped me around to saying that XML was a downright
bad idea for typed information, but at some point it dawned on me that
the PSVI is really enormous when exposed as XML, and that maybe markup
structures themselves aren't particularly good at conveying the type of
information people want to work with. (W3C XML Schema's a separate but
similar story.)
Basically, it's great that we have all this metadata around describing
how best to strip text down to a lot smaller set of bytes so that
programmers can actually work with it instead of thinking about it, but
didn't they want the bits anyway?
Maybe it's too much of a story, but I think it's time to seriously
consider where we've come from and where we're going.
Text was a critical ingredient in convincing the world that this kind of
information interchange was _possible_, but I'm not sure that the same
kind of convincing is necessary now, buzzword-compatibility aside. It's
at least worth some serious consideration, preferably before the world
wakes up to this.
--
Simon St.Laurent
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com
|
