Lists Home |
Date Index |
- From: Didier PH Martin <firstname.lastname@example.org>
- To: Eugene Kuznetsov <email@example.com>,Eric van der Vlist <firstname.lastname@example.org>, Paul Tchistopolskii <email@example.com>
- Date: Thu, 14 Dec 2000 09:03:13 -0500
> From: Eric van der Vlist [mailto:firstname.lastname@example.org]
> 1) "get the document named 'some.xml'" or
> 2) "get your representation of the document named 'some.xml'" ?
> I personally handle "GET /some.xml" in the second meaning in my sites
> thinking I can give more added value like this...
Doesn't that run against the whole idea of separating content from
presentation on the web at large? (As opposed to separating it
server-side, which most web developers did long before XML).
The underlying problem here is that we have a recommendation stating that we
can include a processing instruction in an xml document. When this
processing instruction is present, an interpreter, for instance, an XSLT
interpreter, can transform this xml document into a new one. This is the
"recommended" way of transforming a model into a view.
I think the whole issue is around the fact that this "transformation" may
occur on the client side or on the server side. In fact, when I dig further,
if process partitioning is allowed, if process partitioning can be turned
Pros of having the "transformation" process to occur on the server side:
a) can adapt to devices having few resources (i.e. memory, processing power,
wrong interpreter, etc...)
b) can perform some load balancing between two processors (the requester and
c) help some business model to adapt to different devices but force to see
the rendered version that includes some ads (may not be an advantage for the
client but is surely a way to feed these people's children :-)
d) help to personalize the content (some call this mass customization,
others, more pragmatic, call this a way to keep the eye balls, others, in
panic mode, call this a way to stay alive :-)
Allows the any device promise.
a) not able to see the xml document source (i.e the original document before
b) may impose too much work on the server if no partitioning on the clients
is possible (this is actually the case if you are using XSLT 1.0 as a
Suggestion to other fellow server providers doing their transformation
process on the server side:
Provide a switch to set to ON or to OFF the transformation process. My
humble suggestion is to use a URL parameter (easy to implement). So if a
user performs a request like:
GET mydir/mydoc.xml?style=no, then the returned document is the original
try the following URLs to see this in action:
Should the server be set with the "transformation" process on the server set
to OFF? not very practical if you access the server with a WAP device.
Anyway, its a question of taste and choice but if you provide a way to set
to OFF the transformation process, people can see the original xml document.
However, your business model may prevent you to do so. In that case, forget
what I said.