Lists Home |
Date Index |
On Tue, 29 Apr 2003 16:56:01 -0700, Joe English <email@example.com>
> Mu. We're talking about "Namespaces in XML"; namespace names
> aren't URIs (or IRIs), they're opaque character strings which
> are supposed to conform to the syntax of URIs (or IRIs).
> The network is irrelevant.
For what very little it's worth, three years of this discussion has simply
convinced me that in general namespace names should be URNs: like namespace
names they're abstract identifiers, and AFAIK there are no issues about
comparing them. Tim Bray and Berners-Lee argue against this in the most
recent TAG IRC:
Bray: I remain convinced that URNs are almost always a bad idea for XML
TimBL: because either they're not resolvable or you're reinventing HTTP.
TimBL's right, of course, but: a) there's still no consensus on what
dereferencing a namespace URI should produce; b) there is a very
significant *conceptual* price people pay, and pay, and pay for getting
mixed up by the different semantics of namespace names and URIs (such as
internationalization/localization and comparison, just to name this week's
hot potatoes); c) the 'unenlightened' but influential folks in Redmond WA
and Washington DC are voting with their feet for URNs and this debate is
likely to become moot Real Soon Now; and d) W3C could have "reinvented
HTTP" to resolve URNs in a distributed, reliable manner several times over
with the effort that has gone into this debate.
Speaking only for myself, and I will now shut up and crawl back under my
rock for another year or so until the permanentness of this permathread
gets under my skin once again :-) I KNOW that there are good arguments
against this rant, but it's just soooooooo frustrating to watch the W3C
"grind to a halt" over this that I just can't think about them anymore.