W3C home > Mailing lists > Public > www-xml-xinclude-comments@w3.org > March 2002

RE: XML Schema for XInclude

From: Jonathan Marsh <jmarsh@microsoft.com>
Date: Tue, 19 Mar 2002 14:43:27 -0800
Message-ID: <330564469BFEC046B84E591EB3D4D59C0554BF22@red-msg-08.redmond.corp.microsoft.com>
To: "Jeni Tennison" <jeni@jenitennison.com>
Cc: <www-xml-xinclude-comments@w3.org>
Let me run your point back in detail to make sure I can communicate this
to the group correctly.

1) XInclude uses the "value" of the parsed attribute.  This should
probably be clarified that it uses the [normalized value] infoset
property.

2) The infoset [normalized value] property refers to the XML 1.0
attribute normalization section.

3) XML 1.0 collapses all whitespace runs into a single space, and then,
if the attribute type is not CDATA, it strips any leading and trailing
spaces.

4) The DTD fragment in the XInclude spec illustrates the parse attribute
as an enumeration - which is not CDATA and thus would strip leading and
trailing spaces.

5) The Schema fragment in the XInclude spec uses an enumeration of
xs:string, which does not strip leading and trailing spaces.

6) So, the DTD and Schema fragments are inconsistent as to whether
leading and trailing spaces are removed or not.  This needs to be fixed,
and xs:token looks like a good candidate fix.

7) In addition, XInclude processors may have an interoperability problem
depending on whether or not a DTD (or Schema) has been applied.  The
attribute parse=" xml " will have a normalized value of "xml" if a DTD
is present, and a normalized value of " xml " if one is not.
Presumably, the latter fails.  The Core WG should assess the impacts of
this problem.

Sound about right?

Thanks
- Jonathan

> -----Original Message-----
> From: Jeni Tennison [mailto:jeni@jenitennison.com]
> Sent: Wednesday, March 13, 2002 1:25 AM
> To: Jonathan Marsh
> Cc: www-xml-xinclude-comments@w3.org
> Subject: Re: XML Schema for XInclude
> 
> Hi Jonathan,
> 
> > As far as whitespace normalization using xs:token, I don't think
> > that's a good idea. The XInclude processor looks for the strings
> > "text" and "xml", and any whitespace will cause it to fail. If
> > xs:token encouraged someone to put whitespace in there, inclusion
> > would only work when the schema was applied and performed
> > normalization. I think it is a better description of the inputs of
> > the XInclude processor to not allow any whitespace.
> 
> I don't think that the fact that XInclude processors don't normalize
> the value of the parse attribute is made explicit anywhere in the
> spec, but OK. Bear in mind that describing parse as an enumerated
> attribute in the DTD has the same implications regarding whitespace as
> basing its value on xs:token in the schema.
> 
> Cheers,
> 
> Jeni
> 
> ---
> Jeni Tennison
> http://www.jenitennison.com/
Received on Tuesday, 19 March 2002 17:44:01 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Thursday, 9 June 2005 12:16:08 GMT