Re: [xml-dev] RE: An inquiry into the nature of XML and how it orients o

[Kurt Cagle <kurt.cagle@gmail.com>]

Elliotte,

The pointer aspect you give here is one of the big hurdles I have when explaining RESTful services to people, but it's a critical distinction that often tends to get lost when discussing modeling, especially since it changes the nature of the conversation fairly dramatically. A collection, in restful terms, is an abstraction that manifests itself as a contained sequence of links (i.e., pointers) with associated metadata. The metadata may include a representation of the data, but that's not necessarily relevant - the links are.

One of the major distinctions between relational database collections and XML ones is in the nature of their pointers. In an RDBMS, such pointers are, perforce, always local, always constrained to the database in question. With XML, on the other hand, the data may very well be external to the database query engine invoking the query. In Roger's bookstore example, for instance, the key notion in both cases is the ISBN, which can effectively serve as a mechanism for formally identifying the book record in question. Note also that the examples discussed previously do not in fact point to specific books, only to records indicating that such a resource is in fact identified with the system; there's no indication whether there are in fact any physical books that match that ISBN, or the number of books that match that ISBN, or even whether such a book is obtainable from within the existing system. 

Not sure where I'm going with this, other than to concur with your comments that pointers play a larger role in the XML ecosystem than many people give much thought to.

Kurt Cagle
Managing Editor
http://xmlToday.org

On Mon, Nov 23, 2009 at 7:37 AM, Elliotte Rusty Harold <elharo@ibiblio.org> wrote:

On Mon, Nov 23, 2009 at 7:46 AM, Costello, Roger L. <costello@mitre.org> wrote:

> The OO form frames ones thoughts in this way:
>
>   Book stores contain a collection of Book objects
>   and Magazine objects. The Book and Magazine types
>   inherit from an abstract Publication type.
>

Sorry but this is wrong. Bookstores do not *contain* a collection of
Book objects. It seems you have now spent so much time with XML, that
XML has colored your thinking to the point where you try to fit
everything into an XML model. What you describe is an XML description.
Don't worry. It happens to everyone. It's just that I'm used to seeing
people try to force XML to fit the OO or relational models instead of
the other way around. :-)

The OO description is that a bookstore contains a *pointer* to a
collection of Book objects. Very rarely would a bookstore object
actually contain the books. The distinction is subtle, but critical
most especially when attempting to serialize object graphs into XML
trees or relational tables. Two different bookstore objects can have
pointers to the same books. However, two different bookstore elements
can't contain the same books. Trees are a restricted subset of graphs,
and this is why naive approaches to serialization are doomed to fail.
(Sophisticated approaches to serialization are also doomed to fail,
but for more complex reasons.)




--
Elliotte Rusty Harold
elharo@ibiblio.org

_______________________________________________________________________

XML-DEV is a publicly archived, unmoderated list hosted by OASIS
to support XML implementation and development. To minimize
spam in the archives, you must subscribe before posting.

[Un]Subscribe/change address: http://www.oasis-open.org/mlmanage/
Or unsubscribe: xml-dev-unsubscribe@lists.xml.org
subscribe: xml-dev-subscribe@lists.xml.org
List archive: http://lists.xml.org/archives/xml-dev/
List Guidelines: http://www.oasis-open.org/maillists/guidelines.php