> > [Not a grove at this stage, as
> > no one seems to write their parsers to create groves.]
And also, a GROVE is more a modeling tool to allow clear
explaination of how, for example, tree addressing works: questions
like "when an element contains an attribute, is the
element the 'parent' of the attribute?"
It also fills in the gap in the SGML spec "what does
a parser (nominally) actually find in a document?"
and also answers "what objects are there for DSSSL
to manipulate from an SGML parse".
In the same way as the OSI networking model is very
good for teaching and explaining networking concepts,
but is not so good for implementations and does not
actually model the way the most successful networking
(i.e. TCP/IP) works, so the GROVE model may be successful
even if it were inapropriate for XML parser-writers to
use in designs directly.
Rick Jelliffe