Re: XPath and a continuous, uniform information space - Recap

On Sat, 17 Aug 2013 23:00:46 +0100 (BST), Hans-Juergen Rennau <hrennau@yahoo.de> wrote:

>Jeremy, thank you very much - this sounds most interesting. Could it be that that the designers of XPath/XQuery/XSLT might assimilate and possibly generalize some of the concepts which have been developed for the dynamic construction of technical documentation - a construction which, abstractly speaking, can be regarded as the imposing of secondary structures on a primary node forest? It seems to me a perfectly reasonable idea. Would you agree that it is at least conceivable?

Sure.  In fact, this part of the discussion puzzled me;
DITA is one of the more popular uses of XML, in pubs at
least, but nobody made the connection to mapping.

Regarding having the nodes contain their own maps...
the equivalent in DITA is inline <xref>, which is
discouraged, and <related-topics> which is tolerated,
but for which keeping the intertopic relationships
in the map instead (in a <reltable>) is preferred.
The concern is that either construct creates an
interdependency between nodes, so that re-use of
one node but not of the others it references is
either impossible (<xref>; broken links) or is
ambiguous (<related-topics>; if not in the map,
what do you do about the link?).

This is partially ameliorated by the indirect-
addressing method (@keyref, <keydef>) where you
can have different destinations for the same xref
when the doc is in different maps, by including
an appropriate key definition in the map itself.

I'm interested to see where the folks here take
these considerations when looking at them afresh.


-- Jeremy H. Griffith <jeremy@omsys.com>
  DITA2Go site:  http://www.dita2go.com/