Re: Poll: Relative URIs and Strings in xmlns attributes

I agree both with Martin's conclusion and reasoning.

The other choice leaves open the possibility that at some future point
the DSig group will be left with the unpalatable choice of being
incompatible with a previous version or incompatible with the rest of
the W3C XML Recommendations.

"Martin J. Duerst" wrote:
> 
> I very clearly prefer (1), because it's fully aligned with the
> rest of the XML world. If it becomes clear how to handle relative
> URIs in namespaces, we can update, with an update of the relevant
> URIs so that it's always clear what's intended.
> 
> The argument that Canonicalization is URI-unaware doesn't hold,
> because it can't be namespace-aware and URI-unaware.