Re: layering is consistent and coherent

On Sat, 3 Jun 2000, Al Gilman wrote:
> Namespaces, per se, don't posess identity.  

As I remember, "identity" is the only operation
which the specifification attempts to define.

> There is no sufficiently useful and general equivalence 
> test for namespaces by which to define namespace identity.

No doubt, that some think this is the reason why we are here.

> Identity for individual namespaces will be found in subclasses 
> of namespaces, not in the namespace domain itself. 

Ouch!

> The XSLT namespace is an example of this.

Yes, it defines namespace comparison requiring absolutization
first.  What happens if my XSLT processor receives its XML 
document SAX stream from a parser which does not absolutize
and does not provide the "base:uri" ?   It is not a compliant
XSLT processor.  This to me is unacceptable.  Namespace 
processing should be done *once* and hopefully in the parser.

...

Namespaces should be considered, as you brilliantly 
suggest, as "marks" -- which by their nature are 
both a singleton and unique for a given domain.

I had a day dream after reading your other post a while
back.  The dream had a bunch of cattle, each had a name 
tag, and each had a brand burnt into their hide.  They were 
moving into a coral area and were being sorted by a 
fella called "XPath".   The last thing I want is the 
branding of my cattle to be in-flux among specifications.


Best Wishes,

Clark