Article:		XLink: Who Cares?
Subject:		missing pieces
Date:		2002-03-18 23:17:14
From:		Erik Wilde
hello. in my opinion, one of the most important reasons for the lack of success of xlink are the missing pieces. in the same way as xml lacked a proper data model before the infoset came along, we also need an xlink infoset. the w3c note on xlink and styling has made a frist attempt, but it has holes. until there is no formal definition of the xlink infoset, it will be hard to reasonably define things such as xlink transformation, linkbase accesses, queries into the linking structure and similar things. i guess we can learn from history here, which has shown that as soon as you want to do non-trivial things with xml (such as xslt or xquery), it is absolutely necessary to have a data model. another thing is that software companies will not really push things like xlink, which basically is about an open web. when you look at microsoft's smart tags, you can see that what they are doing is nothing more than generic linking under the control of microsoft, which is much more in their interest than general xlink support. general xlink support in browsers (given a suitable linkbase access method) would enable generic linking on the web for everyone.

Previous Message

Next Message

• missing pieces
  2002-03-19 22:00:17 Adrian Edwards [Reply]
  
  Not sure that lack of a data model is the real reason behind the icy community response. The HyTime standard (ISO/IEC 10744:1992), which included linking formalisms for SGML that parallel XLink's for XML, added a very rigorous data model known as the ‘grove paradigm’ [1] with its Second Edition (ISO/IEC 10744:1997), yet has suffered a similar (and similarly undeserving IMHO) fate. Perhaps we need to look elsewhere to identify why XLink has not caught on.

  
  I believe that the most significant contributing factor in XLink's stagnation is the lack of an evangelical core community that Bob rightly highlights in his article. Is it a coincidence that the most recently updated page on the http://www.hytime.org site that Bob links is 31 August 1999? No. The key intellectual figures in generalised hypertext markup in the nineties: Newcomb, Kimber, DeRose, Prescod and Biezunski, celebrated the new millennium by forming their own consulting companies and abandoning HyTime and XLink on the speaking circuit. Those most capable of playing the role for XLink that say

  
  It is also no coincidence that these figures (Newcomb and Biezunski in particular) are now largely responsible for the "almost exasperating" current level of exposure for Topic Maps at industry conferences. Where would XLink be if the same energy and experience were brought to bear evangelising its virtues? Perhaps in no better position than HyTime, although the relative simplicity of XLink does give one some hope that it could appeal to the mass market, given the right publicity.

  
  Perhaps however, as admitted in XLink’s own 'Design Principles' [2] "... certain functionality, in particular out-of-line link handling with extended document groups, is inherently difficult."

  
  [1] http://www.prescod.net/groves/shorttut/
  [3] http://www.w3.org/TR/NOTE-xlink-principles#1.9