Re: The questions and their relationship

At 15:01 2000 05 18 -0400, Paul W. Abrahams wrote:
>2. When comparing URI's for equality, should the comparison be based on their
>literal string representation, possibly after some simple normalizations, or
>should it be based in some way (possibly using absolutization for relative
>URI's) on what resource those URI's refer to?   (Different specs appear to
>suggest different answers.)

I really do not think anyone involved in the XML Plenary or earlier
XML Core or earlier XML WG discussions has ever seriously suggested
that namespace name equality be based on anything that required
internet access.  I think it would be productive if we all agreed
to stop bringing up de-referencing (i.e., accessing the resource)
as a possible part of namespace comparison.  That is NOT what anyone
means by "absolutization" or even "canonicalization" in this discussion.

Among the choices discussed in the XML Plenary straw poll were
literal string comparison of the values as given in the namespace
declarations, comparison on the string values after being absolutized
per the algorithm in RFC 2396 (as augmented by XML Base), and 
comparison on the string values after being absolutized and 
canonicalized further (via some not as yet precisely defined 
algorithm).  We never suggested de-referencing would be part of 
namespace name comparison.

>3. Should the URI reference in a namespace attribute point to a resource of
>some kind that at least hints at (and perhaps defines completely) the meaning
>of the names used within that namespace, or should it have only the
>significance of a unique identifier?   The Namespace Specification
>categorically says that its only significance is that of a unique identifier,
>but TimBL has suggested that this answer may be the wrong one.

This issue is discussed in relationship to namespace names because it
apparently affects the way people think of what a namespace name is.
But, as you point out, this issue is separate from that of how one
compares namespace names.

paul

Received on Thursday, 18 May 2000 16:02:40 UTC