A corollar of having multiple roots in one document is to define a
syntax for naming multiple father-children relations.
As with a database, each element should be accessible directly
(efficient use of '//element') while oriented relations (child to father
or father to children) are used to access other elements.
With XML 1.0, we have to add id attributes to implement this and there
is no specific syntax. XML 2.0 should develop a richer syntax for axes.
XML 2.0 could then be able to represent effective databases (if not that
big) and optimization could be done to manipulate data in memory with
maximal efficiency (pointers and so on...)
Alain COUTHURES
<agenceXML>
