If browsers supported XML + CSS well,
consistently, robustly, then you might see more XML on the web (but see
#2). Browers support (x)HTML, so the web gets XML + XSLT = (x)HTML.
This is less about XML and more about what browsers support.
The problem with XML + CSS is that you very basic “data
model” to rendition. I guess in, that “lack of support”
statement I guess you refer to XSLT 2.0 since XSLT 1.0 is supported by more
than 94% of the browsers on the market. I sincerely doubt very much that the
lack of XML on the visible web is caused by a lack of capabilities in browsers.
It’s more a question of copyright, business model, etc.. not really
The distinction made between "visible
web" and "hidden web" seems to be made based on
"human" versus "machine" consumption. This
distinction may be misplaed when one considers that XML over HTTP (without "web
services") and web services (i.e., SOAP XML + XML over HTTP) can be
delivered to a desktop application for human consumption. Yes, the XML
goes through an application/machine, but what is a browswer -- it is an
application/machine as well. XML over HTTP allows a developer to break
away from the web browser, which provides consistency, speed, and control that
a browser does not provide. XML over HTTP allows one to avoid the browser
and communicate with a desktop application that, in my view, makes human
consumption much better/easier.
What Roger meant by “machine”
can also be translated by any other agent other than the browser. It
could eventually be translated into an interactive view presented to users but
the source code of that application is probably not encoded into an explicit
model to model transformation as we would do with data encode into xml and a
rendering language based on XML. For instance, the transformation process into
something the users can manipulate can be encoded into a Java application. When
the latter is compiled, try to find the linkage or transformation between the
data model and the final rendering. With an XSLT transformation you have a
white box able to show you the relationship between the data model and the
result of the transformation. Most of the other XML based apps are mostly based
on a black box model where the transformation is convoluted into procedural languages
and made opaque after a compilation. It is also the case that most of these
apps are more of a transactional nature than for user manipulation.
Didier PH Martin