From: Danny Wyatt (danny@cs.washington.edu)
Date: Tue Apr 13 2004 - 23:34:19 PDT
The authors present a method for validating XML documents against
schemas that amounts to type checking the elements and atomic values in
the document. If an element matches---via named types, derived types,
or normal value to type matching---a type *and* it can be converted to
that type and re-serialized to its textual XML representation, then it
is a valid instance of that type.
Perhaps fortunately, I have not had to read the XML Schema spec. As
such, much of the type system in this paper seems intuitive or
obvious---as the authors admit. The "double checking" of roundtripping
is a thoughtful touch for a representation designed first as a data
serialization scheme. The only problem they have with ambiguous types
is due to a feature that should probably never have been included in XML
Schema to begin with. Lists can be represented as a series of elements,
so there is no need to represent them in strings within an element or
attribute. There should have been something analogous to first normal
from for XML that prevented that.
However, that feature does make sense when XML is seen as a
human-typable markup language. This is behind much of the difficulty
with applying a formal type system to XML. There do still exist many
uses of XML for the "light" marking up of text documents. These may not
benefit as much as the more structured data interchange uses of XML from
a rigid type system---and the complex queries it exists to support.
This archive was generated by hypermail 2.1.6 : Tue Apr 13 2004 - 23:34:25 PDT