OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: Why the Infoset?

[ Lists Home | Date Index | Thread Index ]
  • From: "Simon St.Laurent" <simonstl@simonstl.com>
  • To: Paul Grosso <pgrosso@arbortext.com>, xml-dev@lists.xml.org
  • Date: Mon, 31 Jul 2000 19:02:56 -0400

At 05:17 PM 7/31/00 -0500, Paul Grosso wrote:
>At 17:57 2000 07 31 -0400, Simon St.Laurent wrote:
>>...I don't find
>>the Infoset adequate either.  Its philosophy of partial abstraction (at
>>least it's not meta-meta-abstraction) just seems plain broke.
>It might help me to understand your position better if you
>could compare and contrast your feelings of the Infoset [1]
>with the XPath data model [2].

There's no need for contrast - I find them both lacking in pretty much
similar ways.

Both present a 'processed' view of a document, with much of the information
that led to that processed view - from entities to DTDs to character
references - stripped out entirely.  There's no notion that something which
has been defined and understood has been lost, or that those somethings
might themselves have worth in certain contexts that make them worth

In the case of XPath, I find such reductionism acceptable - I'm not
convinced that preserving it is worthwhile in the contexts where XPath is

In the case of the Infoset, I find the reductionism distressing.  Although
the specification certain acknowledges that something is lost, it also
describes itself as:

>a description of the information available in a well-formed XML document

That's fine - except that the Infoset is the only 'canonical' description
available, and a large chunk of information is excluded from the Infoset,
presumably never to return. 

I'd have fewer objections to an Infoset that described all the parts of XML
and then described a process whereby those parts are combined into a subset
that is the view presented by the Infoset, but I find the current form of
the Infoset misleading at worst, an extra layer of vocabulary at best.

I don't mind simplifying XML - that's for certain.  But that doesn't seem
an appropriate task for the Infoset, which I feel strongly should reflect
XML 1.0 and namespaces, as they are.

>[1] http://www.w3.org/TR/2000/WD-xml-infoset-20000726
>[2] http://www.w3.org/TR/xpath#data-model

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
http://www.simonstl.com - XML essays and books


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS