- From: Dan Brickley <danbri@w3.org>
- Date: Wed, 23 Apr 2003 05:41:30 -0400
- To: Joshua Allen <joshuaa@microsoft.com>
- Cc: "Roy T. Fielding" <fielding@apache.org>, pat hayes <phayes@ai.uwf.edu>, uri@w3c.org
* Joshua Allen <joshuaa@microsoft.com> [2003-04-22 17:11-0700] > > What is wrong with saying: > > A resource can be anything that can be named. > > It appears that is what you mean to say That has two readings: - could in principle be assigned a name - could be mentioned by name (because it has one) The former gets us back into resources are just things territory, the latter view, roughly the "to be is to be the value of a uri" view of 2396 resources, is one I find quite worrying. As Pat noted just now, it is a coherent view but one with significant implementation and specificiation costs associated with it. I don't really want to have to keep a timeline for each resource-or-thing to keep track of which things currently "count" as resources. Not least since there is no practical way of telling at any point in time which things in the world have URIs that denote them. For example, I don't currently believe any URIs denote me, but nor do I care or for that matter know how I'd find out. But there is a truckload of RDF software out there that just gets on with the job regardless, using reference-by-description strategies and other conventions for making reference to me without using URIs. Right now that RDF code will treat me as a 2396 resource, just as it treats my homepage and mailbox as resources. Do we have any practical reason for asking that RDF and other Web data systems start to distinguish things that "can be named" (in the 2nd sense above) from those that can't? If not, then let's run with "could in principle be assigned a name", which boils down to "resources are just things". (Hmm so long as we don't actually try to assign all those names in reality, but that's another story.) Dan
Received on Wednesday, 23 April 2003 05:41:33 UTC