XML Schema WG comments on XLink and XBase

In response to the request for last-call review of the XML Linking
Language (XLink) and XML Base specifications, the W3C XML Schema
Working Group recently appointed a task force to prepare comments on
these specifications, for action by the WG.  At its meeting today (17
March 2000), the XML Schema WG discussed these specs and our comments
on them, which I am pleased to transmit to you herewith.

Owing to the pressure of time, the WG did not reach conclusive
agreement on some questions of wording. I have therefore included in
the text below alternative formulations for some points the WG would
like to make.  As will be clear, some members of the WG wished the
comments to be more specific in what is proposed, while others
preferred to state general points and leave the details to the WG.

The WG as a whole did agree that an XML Schema for XLink would be
desirable, and that XBase information must be available in the
information set, in order that XML Schema processors, and other
processors with specialized knowledge, can process URI references
properly even when generic XML processors cannot reliably identify the
values in question as URI references subject to XML Base resolution.


-C. M. Sperberg-McQueen
 Co-chair, W3C XML Schema Working Group

---------------------------------------------------------

The XML Schema WG would like to congratulate the Linking WG on
reaching this advanced state of standards evolution.

XLink

The Schema WG believes that the specification would be stronger and
XLink applications would be more interoperable if a XML Schema for
Linking were included in the specification.  Since this would not
change the XML document syntax nor the processing implementations we
request that this be done during the Candidate Recommendation period.

XBase

XML Schema is an application that needs to use xml:base declarations
in conjunction with both schema constructs that carry URIReferences
(which may be relative) as attribute values, and with user data of
URIReference datatype.

[There was some disagreement over the best wording for the
next paragraph.  Three alternative wordings are available, and
convey the range of opinions in the XML Schema WG. -CMSMcQ]

[1] XML Schema is therefore dependent on the Infoset to carry full
information about xml:base declarations, preferably as information
items to reflect the similarity of their scoped usage to that of
namespace declarations. In this way, the Schema language and other
applications will be able to interpret relative URIReferences
occurring as attribute values or in text content consistently with the
scope of xml:base declarations, if they so choose.

[2] XML Schema is therefore dependent on the Infoset to carry full
information about xml:base declarations.  This should preferably be as
core properties on all relevant information item(s) sufficient to
enable convenient determination, for any element in the infoset, of
the in-scope xmlbase value (if any). In this way, the Schema language
and other applications will be able to interpret relative
URIReferences occurring as attribute values or in text content
consistently with the scope of xml:base declarations, in any situation
in which xml:base may have been used.

[3] XML Schema is therefore dependent on the Infoset to carry full
information about xml:base declarations.  This should preferably be as
core (required) properties attached to element information item(s)
sufficient to enable convenient determination, for any element in the
infoset, of the in-scope xml base value (if any). Specifically, we
propose that the base URI attribute be provided as specified in the
XML Base draft [1], but that it be made mandatory (core), not optional
(peripheral).  In this way, the Schema language and other applications
will be able to interpret relative URIReferences occurring as
attribute values or in text content consistently with the scope of
xml:base declarations, in any situation in which xml:base may have
been used.


-- 
****************************************************
* C. M. Sperberg-McQueen                           *
* Research Staff, World Wide Web Consortium        *
* Route 1, Box 380A, Española NM 87532-9765 *
* (that's Espanola with an n-tilde)                *
* cmsmcq@acm.org, fax: +1 (505) 747-1424           *
****************************************************

Received on Friday, 17 March 2000 14:33:01 UTC