Re: Use cases

At 11:25 AM 2000-05-22 -0400, John Cowan wrote:
>David Carlisle wrote:
>
>> Namespaces are defined to be URI references, with an explicit
>> character-for-character equality test.
>
>Actually, it is namespace *names* that are defined to be URI references.
>But this does not affect your argument.

In constructing the [shrouded in Member space] straw poll, the chairs took
pains to avoid acting as though the Namespaces Rec defines what one should
call a namespace *name*.

Would you consider that excessive caution, or well chosen art in the art of
searching for consensus?

To me, it is helpful to say that the "ns-attr" in the namespace declaration
is clearly and normatively required to conform to URI-reference syntax, and
that in enforcing the "no repeated attributes" clause it is to be compared
using literal string comparison.

The "not a goal" disclaimer tries to say that is all there is to write, but
it itself is heuristic and subject to materially important divergence in
reading.

The interpretation that a URI-reference would always, or commonly, be a
reference to a resource by a URI for that resource is not surprising, but
clearly not _implied_ by a close reading of the document.  Could be
inferred not from the document itself but from the larger context of Web
practice.  [Refer to Tim's remarks about "I'm too Web-centric..."]

Unfortunately, the simple terminology "the namespace *name*" fosters
disagreement, not agreement, at this point.  If, following Simon's
suggestion, we want to talk in layered terms, it should be understood that
the natural home layer of the term 'name' is ambiguous enough to cause
problems.  For the layerwise analysis we would do better to use a less
simple but more clear way to refer to the information passing through a
specific inter-layer interface.

Al

Received on Monday, 22 May 2000 15:37:54 UTC