[
Lists Home |
Date Index |
Thread Index
]
Hi James,
James said:
Some of the ideas remind me of my "Plan X" idea that I mentioned a couple
of times on xml-dev back in 1999:
Didier replies:
Yes James I remember this was an excellent and inspiring post.
Like always, ideas take a long time to percolate in public's mind. Some
ideas out of Xerox PARC (arca 1978) are not even in public's mind yet, so...
Thinking about this topic, it strikes me that Xpath and Xquery are somewhat
implementing part of these ideas. When I use the "document" function in an
XSLT document, then, I am referring to a file as an infoset not as a blob.
Just take as an example the following expression:
http://mydomain.com/linkbase/netfolder/resource[id='this_very_one']
The first part of the xpath expression is saying how to reach the file and
the last part, what element (infoset) I want to extract from the document.
In reality, I think that Xpath is even more powerful than plan 9 expressions
since, it allows custom made infosets not solely restricted by the internal
structure of the document. XQuery is even more powerful but not as well
integrated with the actual HTTP name space.
Digging more, Xpath is the result of the collaboration between an HTTP
server and the XSLT processor since the complete expression is partly
resolved on the server and partly on the client. We are facing, in that case
a balanced API where the workload is shared both by the server and the
client. And in these days of "the return of the mighty mainframe", it is
refreshing to see some balanced processing using the extra power left at the
client side (Sorry, I always have been repulsed by wasted resources - seeing
these gigHz left unused at the client side just remind me that developers
are not always demonstrating their above average QIs :-) Xquery is a little
bit trickier and most of the Xquery processing will most likely occurs on
the server side. Therefore this API is less balanced than Xpath unless we
see some client side implementations, but I doubt this will happen.
However, even if, when seen from the internals, the Xpath expression
resolution is happening both on the server and on the client, from the
developers' point of view, it simply represents a name space. Xquery is more
like SQL and looks more like a query than a namespace.
Finally, I forgot to talk about Xpointer but it seems that this latter is
absorbed by Xpath, so let's forget it.
In conclusion, if we were to speak about the XML framework as an XML based
operating system, we would say:
- OS have always pet or preferred languages, it that case we can say its
XSLT. And XSLT as today as a privileged way to get data XPath. In the next
version Xpath and Xquery but not SOAP. Thus cannot yet obtain data from
services nor even trigger a service from an XSLT document.
- OS have APIs to call for services or get data. In that case we can say
that SOAP seems to be the privileged API with a silver spoon in the mouth
(raised by very rich parents :-). This latter is used to call services.
XQuery and Xpath used to get data (or infosets).
You know, maybe the XML framework is slowly becoming an XML OS.
OK its early in the morning and maybe I am still dreaming :-)
cheers
-------------------------------------------------------------
Didier PH Martin http://didier-martin.com
OpenJade Project http://dsssl.netfolder.com
Coming soon.... http://xml.netfolder.com
Coming soon.... http://blog.didier-martin.com
-------------------------------------------------------------
|
