RE: Personal view

> Is there any use case for this interpretation of relative names?
> It would make it highly dangerous to "repeat" a namespace declaration,
> either explicitly or by making  such declarations be default via an
> attribute default, as while the XML 1.0 attribute is copied, the
> namespace name will change at various points in the document.
> 
> Also, the behaviour of
> 
> <a xmlns="relative">
>   <a xmlns="relative">
> 
> would be completely different to the behaviour of
> 
> <a xmlns="relative/">
>   <a xmlns="relative/">
> 
> and thus it's just something else for a processor to check with no
> benefit (or at most cosmetic benefit) to authors of documents, who
> could more usefully have put the absolute URI explictly in the document
> which would have made it a lot clearer that in the first case both
> elements were in 'xmlns-base:/relative' but in the second case
> the two elements were in 'xmlns-base:/relative/' and
> 'xmlns-base:/relative/relative/' respectively.

I was considering consistency with XML Base. I agree that these examples
are problematic, but if we believe that the results are unworkable for
namespaces, are they not also unworkable for other URI uses?
That is, would you vote against having relative URIs in xml:base
sections apply to embedded URIs as they do in http://www.w3.org/TR/xmlbase
section 4?

Received on Sunday, 18 June 2000 14:07:44 UTC