W3C home > Mailing lists > Public > www-xml-linking-comments@w3.org > January to March 2000

DOM WG comments on XLink Last Call draft

From: Lauren Wood <lauren@sqwest.bc.ca>
Date: Mon, 20 Mar 2000 21:26:59 -0800
Message-Id: <200003210529.VAA00515@mail.sqwest.bc.ca>
To: www-xml-linking-comments@w3.org
The DOM WG is concerned at the usage of qualified names in 
attribute values. The DOM Level 2 isn't aware of qualified names in 
attribute values, and doesn't do fix-ups, so won't help an XLink 
processor. Since the DOM Level 2 does keep all the prefix and 
namespace URI information, an XLink processor does have 
everything necessary for processing, but the DOM WG thinks this 
extension to the namespaces specification (where the possibility of 
there being namespaces in attribute values is not mentioned) could 
cause problems. It's not only the DOM that has problems with this; 
canonicalization and signed XML documents will change the prefix, 
since canonicalization defines what prefixes should be used.

In 3.1.3 XLink uses a URI/local name pair, without specifying the 
syntax, and there is a possibility that the prefix was meant, rather 
than a namespace URI/local name pair. This should be clarified. 
Given the way the DOM models namespaces, using URI and local 
name actually makes more sense (where possible) than using a 

3.1.5 - it appears the entire (potentially large) document must be 
searched to resolve XLinks (although DTD declarations could help).
This could be expensive in a large document, no matter how 
efficiently implemented. We suggest some guidance to users 
for large documents, such as encouragement to declare XLink as 
early in the document as possible. This applies to both external 
linksets and out-of-line links.

3.6.1 - Prefix-in-attribute-value; the "show" attribute contains a 
QName, whose prefix must be declared but must not reference 
XLink's namespace.  This is a specific instance of our concerns 
with prefixes and namespace. There also seems to be a descriptive 
conflict here when it says that other values "must" be interpreted 
as "undefined".

3.6.1. embed: - there is no way in the DOM currently to go from the 
document to the embedded document. This would be left to the 
XLink implementation. 


Lauren Wood, Chair, W3C DOM WG
Received on Tuesday, 21 March 2000 00:29:50 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:32:21 UTC