Lists Home |
Date Index |
Looking in my notes and what I learned about meta information in the last
.... years it appears, at least in my humble opinion, that XMLification
mixed with object inheritance leads to some cool and practical side effects.
For instance, we can say that most of the topic or concept relationships are
based on the notion of set. A topic being a set of resource or, said
differently, that a set of resources can be categorized under a particular
To build these topic sets, the first concept that comes to my mind is the
concept of links or one-to-many links. Also, since a topic can be associated
to a set of properties, then the concept of record ( a la RDBM - or schema
based records) or the concept of frame (free form record or without a fix
schema) comes to my mind.
Playing since already several years with xlinks I found cool and practical
the notion that an element could be used to encode a particular
concept/object and inherit the behavior of a link and thus also encode the
notion of linkage of of some relationhip with external objects. Simple
enough to do in the case of xlink since the element becomes a link or
becomes of type "link" with an xlink:type attribute. Then the element being
a serialized or encoded view of my object/concept _is_a_ link and therefore
inherits the characteristics of a link. Did you know that the concept of
property inheritance seems to be innate to our human brain?
Pursuing this thread, I also found cool and practical to be able to inherit
the frame capacities of RDF. Let's pretend for a moment that it would be
like xlink and that by declaring rdf:type="description" my element becomes
an rdf like element. Thus, my element is now inheriting from xlink
characteristics and rdf characteristics and therefore is_a link and is_a RDF
Playing with this idea (not a new one just one outside of the "official XML
orthodoxy"), I found that I obtained the following gains:
- Being able to re-use stylesheets designed for generic xlink or RDF+ (let's
call it that way since I have no names for it).
- Being able to perform client side transformations and thus reducing the
work load imposed on the server. Moreover, to be able to really do some
process partitioning and fulfill the promises of distributed processing
architectures. Thanks god, maybe we'll move out of the mainframe mind set
again. You know it is tough having lived though a mainframe mindset two
times in a life span. The central bureau concept (my own appreciation of the
mainframe mindset :-) reminds me of Nietzsche's eternal return. (have you
noticed that the number of browsers able to perform client side XSLT
transformation significantly increased in the last year?)
In fact implementing MCF project X with today's XML technologies is a lot
easier. The limitations now are more in the minds than in the tools :-)
Having said all that (sorry if I was verbose but you seemed to have wake up
some forgotten thoughts :-) Micah I have a question for you.
Having a CSS based frame system or RDF like if you prefer has the inherent
disadvantage that no tools are available for rendition. So to speak to
transform the frames into something appearing in my browser. Have you
anything in mind about this issue? You know, with MCF we had the
disadvantage of having to write our own viewers/parsers.