[xml-h] Belated introduction: Indexing

Gavin Thomas Nicol gtn at rbii.com
Sat Feb 22 18:13:50 GMT 2003


On Tuesday 11 February 2003 09:53 am, John Cowan wrote:
> I'm not sure why you say that LMNL doesn't hit the sweet spot for indexing:
> overlapping markup is what it does very well.  What it doesn't do so well,
> at least in the current form, is *structural* markup: in particular,
> [A}[B}foo{B]{A], [A}[B}foo{A]{B], [B}[A}foo{B]{A], and [B}[A}foo{A]{B],
> all mean exactly the same thing: three characters "f", "o", "o" which
> constitute both range A and range B.  To escape this unsatisfying
> conclusion, one has to use layers, which are just as crufty in LMNL as
> milestones are in SGML/XML.

Actually, you can use layers, or you can use the reified layer. In the 
canonical ARA model, the reified layer would be reflected all the way back up 
through nested range constructors so that if you wanted to preserve the 
difference, you could.




More information about the xml-hypertext mailing list