Re: Which QName?

On Friday, August 15, 2003, 3:23:17 PM, Robin wrote:


RB> Hi,

RB> I am unsure that this is an issue that should be considered by the
RB> TAG -- it may be more simply passed on to the XML CG -- however
RB> since the TAG is still working on rdfmsQnameUriMapping-6 and has
RB> resolved qnameAsId-18[0] without a mention that I could find of
RB> this problem, I think the TAG should be aware of the problem
RB> whether it thinks it constitutes an issue or not.

I agree that the TAG should be aware of the issue and that some
guidance should be made available as qnames become more prevalent.

RB> The qnameAsId-18 states that "using the in-scope namespace
RB> bindings has the advantage that it theoretically allows a generic
RB> processor to interpret QNames in content without having to be
RB> aware of any application-specific mechanisms".

RB> That, unfortunately, is very theoretical. There is dissent on
RB> which of the QName resolving rule applies to QNames in content:
RB> for XML Schema's xs:QName, the "element rules" apply so that no
RB> prefix means the default namespace applies; for XSLT, the
RB> "attribute rules" apply, and no prefix means no namespace.

CSS3 also picks the element rules (for both elements and, it seems,
for attributes)
http://www.w3.org/TR/2003/WD-css3-syntax-20030813/#defining

"If the optional namespace prefix is omitted, then the namespace URI
is considered to be the default namespace. The default namespace
applies only to type selectors that have no explicit namespace prefix
declared as described in the Selectors Module [SELECT]. "


RB> I haven't found a document explaining which of those is the best
RB> option, or even putting the emphasis on clearly defining which one
RB> is to be chosen.

The latter, at least, should be possible.


-- 
 Chris                            mailto:chris@w3.org

Received on Sunday, 17 August 2003 10:56:20 UTC