SGML: Groves (Denny-Brown)

Subject: Re: Observations on Seattle Conference

Date: Fri, 30 Aug 1996 19:44:54 -0400

From: Derek Denny-Brown <derdb@techno.com>

Newsgroup: comp.text.sgml



-------------------------------------------------------------------------


[len bullard]
> Eliot, can you comment on how much of this grove stuff 
> would be exposed to an author or should be?

well, I am not Eliot, but I think I am qualified to try an answer this
question...

In general, groves just provide a very precise, structured for for
specifying locations of data in a document.  The "default" HyTime
grove plan for a SGML document will act very much like the ESIS model
used in the original 10744 text.  In fact for most purposes, it will
be exactly the same.  You will still use treeloc to locate in a
document's instance just like before. (If you have looked at the DSSSL
spec you will notice that it views SGML slightly differently.  DSSSL
does not have the concept of a pseudo-element, defined in 8.4.2.2 of
ISO 10744.  HyTime uses a sligntly different grove plan as it's
default, where one of the primary differences from an author's point
of view is the use of pseudo-elements.)

Groves are usefull to an author for 2 primary reasons:
1) They are a specific definition of the structure that is located
   over by the various HyTime and SDQL location queries
2) They provide for a complete specification of a document, allowing
   an author to (potentially) locate ANYTHING in a document, down to a
   record end which was displaced because of a inclusion.  Not every
   implementation need provide all of this detail, and it is expected
   that most will not (well, at least I expect that), but the potential
   is there.

The level to which an author needs to understand groves depends on the
author.  In an ideal world, they would not _need_ to even know they
exist.  It is my opinion that the authoring tools should abstract all
of that.  What is necessary for an author to understand is the various
types of HyTime locators and how they differ.  A good understanding of
the different types of location addressing will allow an author to
write more maintainable documents. (i.e. carefull use of location via
name, with minimal, specific positioning from that name, rather than
giant location addresses rooted at the document element, etc.)


-derek

---------------------------------------------------------------------------
Derek Denny-Brown <derdb@techno.com> | Technical Staff @ TechnoTeacher, Inc.
  http://www.techno.com/~derdb/      |    work-phone: (716) 389-0963
      SGML/HyTime/DSSSL/WWW          |      http://www.techno.com/