W3C home > Mailing lists > Public > www-xml-infoset-comments@w3.org > January to March 2004

RE: Call for Review: XML Information Set Second Edition is a Proposed Edited Recommendation

From: Michael Rys <mrys@microsoft.com>
Date: Fri, 16 Jan 2004 10:07:32 -0800
Message-ID: <EB0A327048144442AFB15FCE18DC96C701BFD4B2@RED-MSG-31.redmond.corp.microsoft.com>
To: "Richard Tobin" <richard@cogsci.ed.ac.uk>
Cc: <w3c-xsl-query@w3.org>, <www-xml-infoset-comments@w3.org>

While the previous version of the Infoset was not calling this behaviour
out explicitly, for the data model generation, the assumption was that
we can assume that the infoset after DTD validation would guarantee

So even if it is not a change in your minds, it is one for people that
read the spec and are not privy to the assumptions of the core WG.

I agree that we can require that the data model generation only
generates from a type-value consistent infoset. 

Paul, Norm: Do you want this filed via public-qt as a last call comment
or can we do it at the F2F next week?

Best regards

> -----Original Message-----
> From: Richard Tobin [mailto:richard@cogsci.ed.ac.uk]
> Sent: Friday, January 16, 2004 6:14 AM
> To: Michael Rys
> Cc: w3c-xsl-query@w3.org; www-xml-infoset-comments@w3.org
> Subject: Re: Call for Review: XML Information Set Second Edition is a
> Proposed Edited Recommendation
> >One of the changes in the Infoset PER has added the last sentence in
> >the following:
> >   [...]
> >   The value of this property is not affected by the validity of
> >   the attribute value.
> >This last sentence is a major type safety problem for
> >XQuery implementations that plan on supporting DTDs based on the
> >optional DTD validated infoset mapping of the XQuery Data Model.
> This is not a change to the Infoset.  It just makes more explicit what
> was already true.  The first sentence of the section described the
> property as
>   An indication of the type declared for this attribute in the DTD.
> without any suggestion that it depended on the attribute being valid
> according to that type.  So unlike the PSVI [type definition] property
> it gives the declared type, not a validated type.
> Note also that attlist declarations (if they are read) are processed
> even by non-validating parsers: attributes are normalized according to
> their declared type regardless of whether they are validated.
> >I therefore would like to recommend changing the last sentence to:
> >"The value of this property is affected by the validity of the
> >attribute value in that it is only set if the value has passed
> >validation."
> That *would* be an incompatible change to the Infoset, and we did not
> intend to make any such changes in the second edition.  All the
> changes are intended to be clarifications or extensions for XML and
> Namespaces 1.1.
> I suggest instead that the XQuery data model only use the infoset's
> [attribute type] value if the attribute is valid.
> -- Richard
Received on Friday, 16 January 2004 13:07:37 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:08:05 UTC