W3C home > Mailing lists > Public > www-tag@w3.org > June 2009

Re: Five mechanical approaches to make an XSD profile without getting bogged by individual issues

From: <noah_mendelsohn@us.ibm.com>
Date: Mon, 1 Jun 2009 23:02:22 -0400
To: Rick Jelliffe <rjelliffe@allette.com.au>
Cc: Paul Downey <paul.downey@bt.com>, "www-tag@w3.org" <www-tag@w3.org>
Message-ID: <OF6F946C66.A429929C-ON852575C9.000FD785-852575C9.0010B3BF@lotus.com>
Rick Jelliffe wrote:

> [Noah Mendelsohn wrote:]
> > * mixed content
> > * <xsd:choice>
> So why are they so sacrosanct that they could not be barred in a 

I'm afraid you missed my point, which was not to claim that they're 
"sacrosanct".  I was reacting to a combination of two points that I had 
perceived you to have made: 1) XSD is an egregiously complex and 
unworkable design; and 2) this is underscored by the workshop in which 
members of the databinding community complained that supporting all of XSD 
was too hard.

Regardless of whether I agree with the points ascribed to you, my point 
was:  the most frequently mentioned problems for the then-current 
databinding implementations were not in general the things that most of us 
would question as excess complexity in XSD, unless you consider choice or 
mixed content to be examples of where XSD went overboard?  Choice and 
mixed content were problematic not because they represent silly complexity 
in XSD, but because a key purpose of XML is to handle document-oriented 
and other data that does not fit into traditional C/Java/(and now JSON) 
datastructures, I.e. simple named fields, with unstructured fields, and no 
tagged unions.  Yes, for some pure data scenarios a "profile" with no 
mixed content (structured strings) or choice (tagged unions), would be 
appropriate, but I certainly would not leave those features out of even a 
small general purpose XSD profile.  Both, of course, have direct analogs 
in DTDs, and having at least the expressive power of DTDs was a 
requirement for XSD, as I recall.

Mike Champion and others have also written that the databinding community 
has now started to do what was proposed as the long term direction for the 
workshop, I.e. to provide support for increasingly more of XSD. 


Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
Received on Tuesday, 2 June 2009 03:03:08 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:33:02 UTC