xml-dev - RE: The Power of Groves

["Didier PH Martin" <martind@netfolder.com>]

Hi Len,

Len said:
Is a grove a means to standardize?  Is it better and why?  For what?
In 50 words or less.

Didier replies:
A Grove's atomic entity is the node. A node is a frame composed of a set of
properties. A property is typed, singleton and collection types are
available.

A frame is like a record but, it can, in contrast to RDB records, contain
collection properties and therefore could easily be used to construct
hierarchies.

Now, let's get practical: Most 3D models are based on the composite pattern
(Gamma & al.) A scene is a whole-part structure, a hierarchy. How a grove
could be used to model such thing?

By creating nodes (i.e. frames) equivalent to object found in the 3d
modeling tool. For instance, a camera object could be translated into a
camera node. This object has a particular property configuration or set of
properties. Thus, the camera node can have a certain property set. If the
camera node is to contain other elements or nodes then its "content"
property would be a collection property.

It is even possible to envision that a scene is to be a recursive structure.
A scene may contain a scene. Therefore a 3D universe may be a recursive
structure of scene. If that is the kind of chosen structure, then a scene
can be translated into a scene node (in Groveland) and each node to have a
"content" property itself a collection of scene nodes. Thus, we have now a
recursive structure or a whole-part structure. Then, a scene can be composed
of...and so on and so forth.

A Grove model is not necessarily constrained to the particular syntax or by
the parsing of this syntax. A Grove model is simply:
nodes: (i.e. frames)
each node contain a set of properties
a property can be either a singleton or a collection.

Sorry I took me 287 words and some redundancy to express the idea. Now, the
best way to decide if Grove are OK for the task or not is through  concrete
experience. The best way is to build a Grove structure for the 3D world and
see if it flies.

Let's take a concrete example, the SoftImage model, it is a model like the
Grove model except that instead of building a hierarchy of object using a
collection property like the kind of property found in Groves, it has the
notion of dual collections (1 - the object collection, 2- the property
collection)

3D SoftImage model:
scene----object---->property set (a set of properties, not the Grove
property set)
    |____ camera --- object---->property set
		etc...

With grove we would have instead:
scene ----properties
      |-- content property
              |___ camera
		  |___ etc...

So, in the SoftImage model you build a hierarchy by having an object to be a
collection of objects. In Groves, you build a hierarchy by having a
"content" property, itself a collection property which contain other objects
and each member of the "content" property can possibly  contain a collection
property named "content". Therefore, the "content" property provides the
recursive construct necessary to build trees.

The best way to know Groves is to use them practically, so, what is the
problem in 3D land that we can try to resolve with Groves. By discussing the
model or using Groves to model it, we will know the difference between
Groves and the monster of the Lock Ness. But actually, it is the same
phenomena, people talk about it but nobody saw it. So, let's show it. Len,
can you express the problem at hand and let's see if this could be modeled
with a Grove model. Everybody will benefit to better know the monster, see
that he's only shy and is, in fact, a good body

Cheers
Didier PH Martin
----------------------------------------------
Email: martind@netfolder.com
Conferences: Web New York (http://www.mfweb.com)
Book to come soon: XML Pro published by Wrox Press
Products: http://www.netfolder.com