- From: Tim Berners-Lee <timbl@w3.org>
- Date: Mon, 29 Oct 2001 18:57:33 -0500
- To: "Harald Tveit Alvestrand" <harald@alvestrand.no>, "Eastlake III Donald-LDE008" <Donald.Eastlake@motorola.com>, "'Rob Lanphier'" <robla@real.com>
- Cc: <uri@w3.org>, "Dan Connolly" <connolly@w3.org>, "Al Gilman" <asgilman@iamdigex.net>, "Larry Masinter" <LMM@acm.org>, "Dan Zigmond" <djz@corp.webtv.net>, "Rich Petke" <rpetke@wcom.net>
One reason you need a mapping between Contentxt-Types and URIs is that one must be able to introduce new non-standard context types with all the benefit of URI machinery - Anyone can make a new one - Choice of schemes with different properties of identity, dereference, etc - Ability to talk about them for example wiht RDF and all other languages which use URIs. There are millions of different sorts of things being registered in URI space all the time many more than IANA could ever cope with. Sean Palmer is right, of course, that any DAML-aware engine can use the unambiguousness of Content-Type to reason about content types. BUT if you use a string, then you have the X-problem again - what do you do about experimental, local use, peer agreement etc Content types? The URI system has all this machinery built itn. It's an hourglass system, like IP. Lots of functionality supporting URIs below the waist, and lots of very different uses above the waist, which specify a URI as the identifier. ----- Original Message ----- From: "Harald Tveit Alvestrand" <harald@alvestrand.no> Sent: Tuesday, September 25, 2001 4:03 PM > Don, > it would be quite useful if your draft would explain better WHY you need a > mapping between these two quite dissimilar name spaces. > > I have trouble imagining the case where you would want to use it. > > Harald
Received on Monday, 29 October 2001 18:57:45 UTC