* Tim Bray <[EMAIL PROTECTED]> [2004-11-10 15:16-0800] > > I've been talking with Scott Hollenbeck, our IETF area director, about > IETF standard practice for namespace names, and he said: > > <scott> > The closest thing we have to a guideline is described in RFC 3688. I'd > recommend following the URN structure described in that document, but > that > leaves some freedom for the string used to identify the namespace. > Something like "atom-1.0" could work, but I'm open to suggestions. > > You can see the list of namespaces that have already been registered > here: > > http://www.iana.org/assignments/xml-registry/ns.html > </scott> > > This would give us a namespace name something like > "urn:ietf:params:xml:ns:atom" which is OK I guess. I said that I liked > HTTP URIs for namespace names because they leave open the door for > putting some helpful info there and for things like RDDL. Furthermore, > in RFC3470, which Scott co-wrote, section 4.9 agrees with me. So do > sections 3.5 and 4.5.4 of the W3C's "Architecture of the World Wide > Web". > > Scott acknowledged this, but told me that if we go with the RFC3688 > flow, there will be no fuss, no muss, and we'll get our namespace name > smoothly and quickly. If we wanted to stand on principle for an HTTP > name, we'd have to explain why to a whole bunch of people, and then > there's the problem that the IETF doesn't maintain web space for this > kind of thing, so maybe it should be in www.iana.org; anyhow, lots of > extra work. > > So... what's the sentiment of the WG?
In the absence of a widely deployed mechanism for getting from urn:* names to information about the things those URIs name, I'd much prefer use of http://* names. But my preferences there tend to be from the RDF world, where I've a better idea what to do with runtime-discovered namespace info. Perhaps not such a big deal for XML-syntax oriented formats? Dan
