Lists Home |
Date Index |
This sounds very interesting to me. I've been thinking along the same lines
for the tool I am experimenting with. Do you have any pointers to resources
on the web discussing any of these ideas?
Currently, the approach I'm thinking of trying is to use a linkbase with
RDDL-like constructs in extended links. Every namespace known by the
workspace would have a corresponding extended link in the linkbase. The
extended link could have just one outbound link to an RDDL resource, but
could also have additional locators and arcs associating the locators with
the namespace. Thus, a user could add additional resources to a namepace
within the workspace beyond those in the RDDL resource on the net (or they
can add resources even in the case where there is no RDDL resource at the
end of the namespace URI).
The other thought I was going to explore is to use extended links to
describe a processing pipeline (e.g. an XSLT transformation). The extended
link would have inbound arcs for the endpoint(s), and outbound arcs for each
resource required for the processing (e.g. source document, template). There
could also be a local resource for template parameters. Within the
workspace, when the user selects an endpoint resource, the workspace matches
it up with the inbound links in the linkbase, and the UI clearly indicates
to the user it is an endpoint and allows the user to refresh its contents.
Thats the idea, anyway. If there are materials discussing such approaches,
or alternative approaches to accomplishing these things, I'd be very
From: Simon St.Laurent
Sent: 1/17/02 6:49 AM
Subject: [xml-dev] RDDL possibilities
If one is willing to accept that there may possibly be value in having
mixed human/machine-readable content at the end of a namespace-URI, RDDL
looks like a pretty good tool.
There was some discussion at XML 2001 about ways to improve RDDL, and
they seemed to come from two related directions:
1) Creating clusters of resources rather than a flat list
2) Describing sequences for resource processing to define pipelines
In some ways these are the same thing, just with sequence mattering in
the second version.
I think it's roughly possible to do these using current RDDL syntax, but
I suspect it might be worth some further exploration of how best to make
these work. Number 2 in particular could take RDDL much deeper into
computer interactions than has typically been the case so far.