Re: FW: W3C XML ID ambiguity

Chris Lilley wrote:

> I don't hear you calling for rescinding xml:space, xml:base and so the
> magic predeclared not remappable namespace code still needs to be in the
> parser, right? So re-using it does not seem to cause additional burden.

In practice, the special purpose code often is not in the 
parser/library. Then again sometimes it is. Often it's just left 
completely unspecified what's supposed to happen. I bring up again the 
case of //*/xml:space and //*/foo:space when xml is not mapped to 
http://www.w3.org/XML/1998/namespace and foo is. Even in the case of XML 
parsing, exactly what sort of error is it when the wrong prefix or wrong 
namespace URI is used here? Parsers/libraries are extremely inconsistent 
on these points.

Furthermore, many libraries do not need to special case xml:space and 
xml:lang at all because they do not provide any API calls that reference 
those characteristics. xml:base is a little trickier because some specs 
like DOM 3 do indeed have a getBaseURL() call or something similar, but 
a lot still don't. With xml:id we're moving deeper into territory where 
libraries actually care about what happens. DOM 2 does have 
getElementById. SAX can return the type of an attribute so it needs to 
recognize xml:id at least for that. XPath also has the notion of IDs, as 
does XPointer's shorthand syntax. xml:lang and xml:space just aren't 
that relevant most of the time, and xml:base only marginally more so. 
However, xml:id is diving right into the heart of a lot of XML 
processing. Any problems that already exist with xml:space, xml:lang, 
and xml:base are going to have much more practical impact with xml:id.

-- 
Elliotte Rusty Harold  elharo@metalab.unc.edu
XML in a Nutshell 3rd Edition Just Published!
http://www.cafeconleche.org/books/xian3/
http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim

Received on Friday, 22 April 2005 22:57:02 UTC