XML Infoset vs XML Tagged Text

[Doug Ransom <Doug.Ransom@xxxxxxxx>]

I think
http://www.imc.org/ietf-xml-use/draft-hollenbeck-ietf-xml-guidelines-04.html
missing the most important point by merely  breezing over XML Infoset in
section 4.4.  I believe the most important use of XML in the IETF guidelines
is the modelling of data as an XML Infoset.  Serialization of XML as tagged
text is of limited value.  
The representation of XML as XML 1.0 Tagged text is not all that important
-- the primary value of tagged text being human readability and reuse of
general purpose off the shelf parsers.  As the guidelines correctly argue,
XML 1.0 Tagged text is not suitable for many applications.

The true value of XML comes in when modelling the data as an XML Infoset.
It makes the design of APIs very simple -- a vendor/designer of a library
that parses and serializes any non-tagged format simply provides access to
the underlying infoset using an existing XML API (like SAX, Java, .NET XML
Reader). This lets
the client of such a library use all their XML processing tools, like XML
Schema or schematron for validation, XSLT and XQUERY for high-level
programming, sax-like libraries 
for streaming/processing of the underlying data, XML Signature and RDF for
analysis. 



The XML Infoset is as important than the serialization.  Different
serializations are emerging that solve many problems.

In my view, the best practices are:
	Use XML Tagged text when practical
	Always define the XML Infoset even when XML Tagged text is not the
serialization format. 


As an example,  an HTTP transaction or a mime message can be represented as
an XML infoset and it would be convenient to access the HTTP transaction
through an XML API.


 



Doug Ransom 
Internet Technologies Domain Lead 
Power Measurement <http://www.pwrm.com> 
<mailto:Doug.Ransom@xxxxxxxx>