groves dissent (was RE: RFC: Attributes and XML-RPC)

Steven R. Newcomb srn at
Fri Sep 24 22:33:55 BST 1999

[Gregg Reynolds:]

> In a word, [the grove paradigm] takes what is basically pretty
> simple - an attributed tree - and turns it into something of
> surpassing obscurity.

You're entitled to your opinion.  How would you prefer to walk
an "attributed tree"?  (And what is an "attributed tree"?)

> But "the grove model" isn't even necessary.  The thing it tries to
> model can be modeled quite adequately without any invented terms or
> concepts.

I don't think there's anything really new in groves, except possibly
the (admittedly terrible) GROVE acronym itself.  How would you model
the thing that groves model?  Remember the requirements: it has to be
a machine-processable formal model, and it has to cover the whole
territory of information resources, not just XML resources, and make
every component addressable.

> But not all great experiments succeed, and groves/DSSSL/Hytime have
> failed in the marketplace for good reasons, and that should guide us
> in building XML.

Let's leave DSSSL out of this.  To lump DSSSL with groves and HyTime
does a disservice to all three.  DSSSL does not describe the grove
paradigm, it only uses it to describe the result of parsing an SGML
document, by means of the SGML Property Set.  If the source of all you
know about groves is the DSSSL standard, then you don't know groves.

And what were those "good reasons" for which groves and HyTime
"failed"?  It looks to me like the primary reasons for HyTime's lack
of mass acceptance, to date, have been the lack of a general toolkit
implementation, and public ignorance of the problem space in which
HyTime offers solutions.  Both of these problems are rapidly being
resolved now, so I think the death-knell of HyTime is still being rung
prematurely.  People ridiculed HyTime's complexity, but then they had
to come up with a way to implement extended linking.  And that led to
the realization that the DOM has no foundation -- it was not, in fact,
an object model.  And that led to the XML infoset activity.  Now W3C
is at the same point the creators of HyTime were, after they
discovered the need for an SGML Property Set, but before they
understood that the ability to express the SGML Property Set depended
on yet another needed invention, an invention that turned out to be
the grove paradigm.  Slowly but inexorably, the XML world is
re-inventing HyTime.

We still don't have an official XML infoset (read: XML Property Set).
We still don't have an official XLink, or an official XML addressing
notation.  Directly comparable features for SGML have all been
standardized in HyTime and in industrial use for years.  While
HyTime's complexity has been castigated by the Web people, HyTime has
been quietly working.  Can it be that the real reason we don't have an
official XLink yet is that the complexity of the real requirements
that XLink must meet is beyond what can be accepted by the Web
community?  Or perhaps it's because XLink, at the power level that
HyTime provides for it, would make RDF a completely unnecessary and
much weaker alternative?  Or perhaps it's because lots of people think
that their own ideas about metadata architecture should become world
standards, instead of allowing metadata architecture to be as
powerful, arbitrary, and flexible as any other XML information
architecture?  That's what HyTime does.


Steven R. Newcomb, President, TechnoTeacher, Inc.
srn at

voice: +1 972 231 4098
fax    +1 972 994 0087
pager (150 characters max): srn-page at

3615 Tanner Lane
Richardson, Texas 75082-2618 USA

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at
Archived as: and on CD-ROM/ISBN 981-02-3594-1
To (un)subscribe, mailto:majordomo at the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at

More information about the Xml-dev mailing list