Re: identity constraints

[Eric van der Vlist <vdv@dyomedea.com>]

James Clark wrote:
> 
> > Recently, I spoke with Jerome Simeon.  I learned that there are three
> approaches.
> >
> > 1) ID/IDERF (as in XML DTD)
> >
> > 2) XPath (as in XML Schema)
> >
> > 3) Types and path expressions (as in UCM)
> >
> > The first approach is very weak.
> 
> Agreed here.

This is one of the key points I have just been covering in the tutorial
I have delivered yesterday (see the URL on my signature)!

I see this as one of the issues with the more general problem of
defining links between nodes in XML documents.

The different approaches that I think we should support as either widely
used or part of W3C specifications are:

1) Application "soft" links (like the example from Murata Makoto with
the company and departments).
2) ID/IDREF "hard links".
3) RDF (that can be handled as 1)).
4) Simple XLink / XPointer
5) Extended XLinks (that can be seen as a combination of 1) when
locators are involved and 4)).

I do also believe that it would be very useful to support the general
case of multi document links, even though the validation would then be
subject to the availability of the resources.

> >The second one is complicated.  In particular,
> > it is to difficult to ensure consistencies of key declarations.
> 
> The XML Schema PR  uses a totally trivial subset of XPath.
> 
> http://www.w3.org/TR/xmlschema-1/#coss-identity-constraint
> 
> This is a major change from previous changes, and makes the XML Schema
> identity constraint mechanism much more reasonable.

On the other hand, the only validation tools that I have found able to
support the whole set of links as defined above are based on XPath
expressions.

With Schematron (and Examplotron that is borrowing Schematron's "assert"
feature), you can define sets of clauses that deal with these issues.

I wonder though if supporting XPath (plus XSLT document() function) that
is probably a hassle for the developers of the validation tools isn't a
great asset for users.

In particular, adding an assert feature can be seen as a "joker" feature
that allows users to define all kind of nasty or user specific tests
that would have been impossible or difficult to implement otherwise.
 
> James

Eric
-- 
See you in Hong Kong for www10:
                     http://www.www10.org/program/w10-half-tut.shtml#ta5
------------------------------------------------------------------------
Eric van der Vlist       http://xmlfr.org            http://dyomedea.com
http://xsltunit.org      http://4xt.org           http://examplotron.org
------------------------------------------------------------------------