- From: Joshua Allen <joshuaa@microsoft.com>
- Date: Mon, 5 Aug 2002 08:21:02 -0700
- To: "Miles Sabin" <miles@milessabin.com>, "WWW-Tag" <www-tag@w3.org>
-1 The first sentence is where things go wrong. URIs on the WWW have never been context-dependent, and never will be. -----Original Message----- From: Miles Sabin [mailto:miles@milessabin.com] Sent: Mon 8/5/2002 2:42 AM To: WWW-Tag Cc: Subject: Re: http-range-14: Suggestions or silence OK, taking http://www.textuality.com/tag/s1.1.html as the starting point, this would be my suggestion for a replacement for the first para, URIs are context dependent identifiers. Typically the context of their use will be shared and implicit, in which case URIs may be treated as if context independent. This cannot, however, be guaranteed in general. In cases where ambiguity is possible and would be harmful, mechanisms which allow independent agents to eliminate ambiguity and coordinate their uses of URIs should be provided. Explicit discussion of all the possible contexts of URI use is out of scope for this document. Whilst not the only framework for the interpretation of URIs, the REST architecture merits particular attention because of its close connection with the underlying protocols and infrastructure of the web. In REST a URI unambiguously identifies a Resource, an abstraction for which there is a conceptual mapping to a (possibly empty) set of representations. The representations of a resource may vary as a function of factors including time, place, and the identity of the agent accessing the resource. For example, at the time this document was drafted, the URI http://weather.yahoo.com/forecast/MXOA0069.html identified a REST Resource corresponding to Yahoo's weather forecast for Oaxaca, Mexico. The representations of this Resource depends on (at least) time, the expressed preference of the user for Fahrenheit or Celsius, and the identity of the user-agent software receiving the representation. The unambiguity of URIs in the REST framework makes it suitable for reasoning about certain classes of automated and semi-automated processing, particularly in cases where either the publisher of a URI is accepted as authoritative or where there is some generally accepted third-party authority. It should be noted, however, that there are cases where these conditions cannot be met and hence where the REST framework of URI interpretation might be inappropriate. RDF, for example, allows weakly coordinated yet interacting agents to make assertions using URIs as identifiers. Here there might not be an authority able to practically rule whether the intended referent of a given use of, for example, http://www.w3.org/, is a particular document or a particular web-site, in which case independent assertions made using that URI might not be directly compatible. In these circumstances agents should be provided with sufficient additional contextual information or disambiguating mechanisms to eliminate any harmful consequences in practice. Cheers, Miles
Received on Monday, 5 August 2002 11:21:37 UTC