- From: <keshlam@us.ibm.com>
- Date: Thu, 8 Jun 2000 15:49:43 -0400
- To: Graham Klyne <GK@Dial.pipex.com>
- cc: Rick JELLIFFE <ricko@geotempo.com>, xml-uri@w3.org
I think the proper reading here is that the URI itself is the namespace's identity. A namespace is defined by a point in URI space. That's _all_ it's defined by. In programming terms: The Namespace occupies an address. We recognize it as a unique object by looking at that address. Nobody has yet defined what data exists at that address, if any. But that isn't a problem for the Namespace spec, which is defined purely in terms of that address. Looking at it this way, part of our debate is over whether it makes sense to point to something which is supposed to be uniquely recognizable by using an offset from the address of our instance document. That may be convenient when you happen to know both addresses and/or can guarantee that the two objects will maintain the same position relationship when moved, but it's a very fragile connection ... perhaps too fragile to serve the intended purposes. ______________________________________ Joe Kesselman / IBM Research
Received on Thursday, 8 June 2000 15:50:00 UTC