[
Lists Home |
Date Index |
Thread Index
]
Dare Obasanjo wrote:
> Before this turns into a pointless critique of the
> XQuery formal semantics I'll just steer this
> conversation back to the original point.
Let's make this simple.
There are instances and classes which are sets of instances. That's all.
Pieces of XML are instances, types are classes.
One can classify an instance using a "classifier". The process of "schema
validation" is simply the process of classifying an instance as or not as, a
member of a class (type).
So a type is a class, and validation is a process of classification.
In any case, you can't have "validation" without some sort of "type" to
validate against ***
Jonathan
*** note that we use the term "type" as a synonym for "class" and not as in
Russell ... A great undergraduate level introduction to logic, which
discusses instances and classes is Tarski
http://www.oup-usa.org/isbn/019504472X.html . It is very simple yet powerful
stuff.
>
> Some members of the XQuery WG feel that a complex type
> system such as XQuery's is necessary for updates while
> I am of the opinion that a post-validation requirement
> is the minimum that should be required.
>
> The bottom line is that any XML database product that
> supports XQuery and W3C XML schema plus an update
> capability will have to support post-update validation
> because the XQuery and W3C XML Schema "type" systems
> leave the door open for too many things to be dynamic.
>
>
> By the way here are some quotes from the Formal
> Semantics document which point out what I mean by
> validation vs. type system
>
> "The XQuery type system is based on XML Schema. XML
> Schema defines a notion of validation for XML
> documents. When doing validation, an XML Schema
> processor checks if the structure, and the text
> content of a document verifies the constraints on the
> structure, and the constraints on values, specified in
> the schema. As an example of constraints on the
> document structure, all book elements might be
> declared to contain an isbn attribute, and a title
> element, followed by a sequence of one or more author
> elements. As an example of constraints on values, the
> isbn attribute might be declared as a subset of the
> XML Schema xsd:string values, using a XML Schema
> pattern facet."
>
> Especially with regards to restriction facets, there
> simply is no way for this to be implemented without
> something akin to post-update validation. Identity
> constraints as well. In fact, when I think about even
> making sure that the results of an update do not
> result in an invalid content model for an element in
> the document is pretty much another validation issue
> not a type one.
>
>
>
|
