W3C home > Mailing lists > Public > www-forms@w3.org > June 2003

Re: The problems with Xlink for integration languages

From: Zarko Berberski <zb_z@yahoo.com>
Date: Thu, 5 Jun 2003 03:38:13 -0700 (PDT)
Message-ID: <20030605103813.72726.qmail@web11002.mail.yahoo.com>
To: www-forms@w3.org

I came across this while searching for more ideas for link classifications and
names for XML tags so was wondering if someone knows of some research/published
work on ways to spell out the kinds of links in tag names - about the opposite
of both Xlink and Hlink hiding them in attributes and URI-s.

For example, Topic Maps spell out that you really have references to
undiscriminated resources (that your system doesn't know what to do with except
show or pass on), references to topics (that your system knows and can process,
analyze, reorganize etc.), and assumed "subject indicators" which your system
trusts as axioms, a basic vocabulary.

But then, with references and ways of their serving being as dynamic as they
are today, they can also be inner (within a current document or topic map),
outer (trusting processing to know where and how to get them), site or system
(trusting  the system above processing to serve) and world wide (trusting the
net or the pure luck). Chances are that there might be a bit more granular or
usable classification with some non-existence rules (for example only topic can
be inner reference).

While it is perfectly fine for a rather low demand processing like a browser to
go parsing URI or pick thorugh several attributes of Xlink, or even element
declaration of Hlink, just about any more formal or deductive (or performance
bound) processing greatly benefits from having a definite and reasonable set of
kinds or references declared by the virtue of tag names. Call them axioms of
references needed in order to make all other automatic reasoning and processing
stable and efficient.

It looks to me like part of this divergence of needs reflects on Xlink, Hlink
and their inadequacy to serve everything from a browser which is perfectly fine
with discerning many kinds by parsing URI, to formalized processing and
reasoning starting with XSLT) which needs at least several kinds spelled out as
axiomatic tags, to systems which need to be able to use essentially keywords in
attributes (which both Xlink and Hlink do) for a third, very different way of

       Zarko Berberski

P.S. Please CC me on any response.

Do you Yahoo!?
Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
Received on Thursday, 5 June 2003 06:42:09 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:36:08 UTC