XML XLink Requirements Version 1.0

After several readings of the XML XLink Requirements document, I'm both
delighted and disturbed.  While I'm pleased to see that several key issues
(4.B.7, 4.B.8, and 4.B.10) have finally been raised within the WG, I'm
concerned that the document as a whole sends XLink down the path blazed by
too many of its predecessor systems: a bold proposal that does too much and
will accomplish too little.

While section 4.A is titled 'General user requirements', the document never
makes clear who these users are.  Instead, the document assumes some kind
of mysterious 'general user' who must have certain needs based on past
experience with previous hypertext systems, as far as I can tell.

But who are these users?

a) hypertext professionals and academics, who've worked with hypertext for
years

b) hypertext software developers

c) hypertext document authors

The general feeling I get from the document is that this requirement set
addresses the needs of a) and to some extent b), but doesn't pay much
attention to the needs of group c).  Like it or not, however, the potential
ubiquity of XLink is directly tied to the reactions of those in group c.
There are many good reasons why most hypertext is authored in HTML today,
rather than Storyspace, Xanadu, InterMedia, or even my old favorite,
HyperCard, and these reasons - simplicity and ease of understanding -
should not be underestimated.

Exploring this document from the perspective of a Web developer, there are
lots of questions about how XLink will work on the Web infrastructure the
W3C oversees.  The future existence of simple links, for instance, is
alluded to only in Section 4.C.3, a not very encouraging next-to-last
place.  The intermediate steps between the 'popular' notion of linking
documents as the creation of traversable paths between points in documents
and the 'extended' notion of linking documents as the creation of sets of
points remain as invisible as ever, and the connections between these two
concepts (except in 4.A.8, which I hope will lead to a formal explication
of these issues) remain buried.

While I realize that many of the people at the W3C and in the XML community
generally regard these specifications as the domain of experts, building
specifications as important as these without providing a clear roadmap for
non-experts is dangerous.  While the previous XLink working draft seemed
workable as it stood in many ways, it was shot through both with large
implementation questions (which paths are traversable?  What does steps
really mean?) and a serious disjunction between 'simple' and 'extended' links.

The XML Linking WG has the opportunity to significantly improve the linking
facilities available to users of all levels by providing a coherent,
usable, and approachable set of tools.  Understanding the audience - and
building tools that are useful to that audience - is a critical step that
shouldn't be ignored in favor of past experience, even when that past
experience has spanned decades.


Simon St.Laurent
XML: A Primer
Sharing Bandwidth / Cookies
http://www.simonstl.com

Received on Monday, 22 March 1999 13:04:57 UTC